Still today, many marketers, finance executives, and leadership teams in companies of all sizes, especially those with a large digital ecosystem, gather together and seem to agree on the following: “We are not comfortable with our online marketing, numbers do not match in the different tools or with our internal data. Measurement is broken and we need to fix it.”

When digging deeper, the main solution that has been driving their digital marketing decisions for the last 15 years is the one they are now trying to fix. They believe there is technical work to do. We must stitch more IDs together, build better models, capture more touchpoints or ignore consent. Then, we will finally be able to scale digital marketing confidently.

But here’s the uncomfortable truth: your attribution measurement is not broken. It’s working exactly as designed. The problem is that it was never designed to tell you whether your marketing is actually working.

The precision trap

The promise of digital marketing was measurement precision. Every click tracked. Every conversion attributed. Every euro accounted for.

This promise created a generation of marketers who believe that if they could just see the full user journey, every touchpoint, every interaction, every click before conversion, they would finally understand what’s working. It even worked really well for growth marketers in some cases.

The result? Companies have spent years building increasingly sophisticated attribution systems:

• Multi-touch attribution models that assign fractional credit across touchpoints

• Data science teams dedicated to stitching IDs across devices and browsers

• Elaborate systems to capture “view-through” conversions with a high dosis of illusion

• Machine learning models trained on post-click behavior

All of this effort to answer one question: “Which of my marketing activities caused this conversion?”

The problem is that this question has a fundamental flaw. Attribution models don’t measure causation. They measure coincidence.

With a limited activity in digital marketing channels, this coincidence can be correlated with causation. However, growing companies get stuck in these measurement systems and they fall in the trap forever. Do you see yourself here? Read on.

The coincidence problem

When a user clicks on an ad and later converts, attribution systems record this as evidence that the ad “worked.” But did it?

Consider the following scenarios:

• A user searches for your brand name, clicks on your paid search ad, and converts. Would they have converted anyway through organic search?

• A user sees your retargeting ad after abandoning a cart, clicks it, and completes the purchase. Were they already coming back to buy?

• A user clicks on a social media ad and converts three days later. Did the ad influence them, or the decision was already made?

In each case, the attribution system faithfully records the conversion and assigns credit. But none of this tells you whether the advertising caused the conversion or merely captured someone who was already going to convert.

Attribution doesn’t measure the impact of advertising. It measures the efficiency of capturing existing demand. How long will you keep making budgeting decisions based on a flawed attribution model?

The Universal Analytics legacy

Much of today’s attribution obsession traces back to a single product: Google Analytics, specially its Universal version.

For over a decade, Google Analytics defined how companies measured digital marketing. It tracked users across sessions, assigned conversions to channels, and produced reports that became the lingua franca of digital marketing. Marketing teams built their entire measurement infrastructure around it.

When Google sunset Universal Analytics in 2023 in favor of Google Analytics 4, something interesting happened. Instead of questioning whether the measurement paradigm itself required a shift and the adoption of new mechanisms, including the modeling capabilities of Google Analytics 4, many companies just wanted continuity and replicated their legacy systems either on top of Google Analytics 4, building their own tracking systems, trying to emulate Universal using Piwik, Amplitude or Mixpanel, ignoring these changes in the ecosystem.

They invested heavily in:

• Customer Data Platforms (CDPs) to stitch identities across touchpoints

• Unclear first-party data strategies to replace third-party cookies

• Server-side tracking to capture what browsers no longer allowed

• Maintaining the Universal Analytics data schema, or adapting Google Analytics 4 to legacy tracking standards

• Complex ID graphs to maintain the illusion of user-level tracking

The goal? Recreate the Universal Analytics world with new tools. Making things worse, those using data from BigQuery (only available in the paid version with Universal, now available also in the free version) tried to replicate the same reports with Google Analytics 4, neglecting the idea that this is not the best dataset for assessing marketing performance anymore. Many are still hitting their heads in the wall with memories of a “better past” and blaming Google Analytics as the source of their problems.

But this misses the fundamental problem. The issue wasn’t that Universal Analytics stopped working, it’s that the approach that Google Analytics popularized will never answer again the questions that matter for growth.

By trying to preserve an outdated measurement paradigm, companies have created increasingly complex systems that:

1. Consume enormous resources in data engineering and infrastructure

2. Focus measurement on demand capture rather than demand creation

3. Create a false sense of precision that doesn’t reflect marketing’s true impact

4. Prevent marketing from becoming a growth engine by optimizing for the wrong metrics

The companies that will thrive aren’t the ones that successfully replicated Universal Analytics metrics. They’re the ones that recognized the paradigm shift and built measurement systems designed for growth, not for tracking.

Why “fixing” attribution makes things worse

The natural response to attribution’s limitations is to try to fix its gaps, include post view paths to conversion, add CRM IDs to increase the cross device coverage. Build better models, capture more data. In some cases, even ignoring consent requirements for the highest goal of having an ID for every user so reports can continue spitting out some data.

But this misses the point entirely.

More sophisticated attribution doesn’t solve the coincidence problem, it obscures it. When your data science team builds an elaborate multi-touch model that assigns 23.7% credit to a display impression and 34.2% to a paid search click, it creates an illusion of precision that doesn’t exist.

Worse, it diverts resources from what actually matters: understanding the true incremental impact of your marketing.

As Les Binet famously said: “You don’t need a tool to measure your marketing impact. You need a science.”

The difference is crucial:

• A tool processes data and produces reports

• A science establishes causal relationships through rigorous methodology

Attribution is a tool. What you need is marketing science, which existed before digital attribution. And it worked.

Before we could track clicks, marketers used:

• Sales data: Did sales increase when we advertised?

• Market research: Did awareness and consideration improve?

• Econometric modeling: What’s the statistical relationship between media spend and business outcomes?

• Experiments: What happens when we turn advertising on or off in different regions?

These methods didn’t require tracking individual users. They didn’t need to stitch IDs across devices. They measured what actually mattered: did the advertising grow the business?

The digital measurement revolution didn’t improve on these methods—it distracted from them. We became so obsessed with tracking individual user journeys that we forgot to ask whether the aggregate investment was working.

The CFO problem

For the past 15 years, marketing teams have been reporting attribution numbers to CFOs, boards, investors, and private equity firms. These reports looked sophisticated: detailed breakdowns by channel, precise ROI calculations, beautiful dashboards.

But they are built on a flawed foundation.

Now, as the limitations of attribution become more apparent via cookie limitations, privacy regulations, platform walled gardens, consent limitations, fragmentation of media and many other reasons, CFOs are asking uncomfortable questions:

• “If our attribution was accurate, why do our MMMs show different results?”

• “If paid search is so efficient, why doesn’t scaling it drive proportional growth?”

• “If our digital ROAS is 5x, why isn’t our business growing faster?”

The reality is that CFOs have been fed a version of marketing performance that systematically:

1. Overvalues lower-funnel activities and remarketing that capture existing demand

2. Undervalues upper-funnel activities that create future demand

3. Ignores the offline impact of digital advertising

4. Misses the counterfactual: what would have happened without the advertising

This isn’t a measurement error. It’s a fundamental misunderstanding of what attribution measures and how marketing decisions should be made.

What you should do instead

If attribution isn’t the answer, what is? Here’s a framework for marketing measurement that actually works:

1. Start with incrementality, not attribution

The fundamental question isn’t “which touchpoints contributed to this conversion?” It’s “how many conversions happened because of my marketing that wouldn’t have happened otherwise?”

This requires experiments:

• Geo-lift tests: Run campaigns in some regions but not others, measure the difference

• Holdout tests: Exclude a portion of your audience from advertising, compare outcomes

• Platform incrementality studies: Use tools like Meta’s or Google’s Conversion and Brand Lift

2. Use Marketing Mix Modeling (MMM) for strategic decisions

MMM doesn’t try to track individual users. It uses aggregate data—media spend, sales, external factors—to estimate the contribution of each marketing channel.

Modern MMM tools like Google’s Meridian make this accessible to more companies than ever.

3. Accept that some things can’t be precisely measured

Not everything that matters can be measured precisely, and not everything that can be measured precisely matters.

Brand building is notoriously difficult to attribute but demonstrably effective. The inability to track it doesn’t mean it doesn’t work—it means your measurement system isn’t capturing what matters.

4. Educate your CFO (and yourself)

The CFO-CMO gap isn’t just about language—it’s about fundamental misunderstandings of what marketing measurement can and cannot tell us.

Have honest conversations about:

• The difference between attributed and incremental value

• Why efficient-looking channels might not be driving growth

• The role of brand building in long-term business value

As Google’s research on CMO-CFO alignment notes: “Effective collaboration between CMOs and CFOs relies on a mutual understanding of each other’s work.”

This is not only about education, this is also about trying to look into the feature without the lens of the past, and let go of company traditions without blaming each other. Admit that the world has become increasingly complex and measurement standards need to adapt. Nobody is wrong, let’s get better together!

The feature, not the bug

The way that cross channel attribution has traditionally worked isn't broken. It's doing exactly what it was designed to do: track post-click user behavior and assign credit to touchpoints.

The problem is that we’ve been using this tool as if it measured marketing effectiveness, when it actually measures something else entirely: the efficiency of capturing people who were already close to converting.

This doesn’t mean tracking and attribution are useless. They serve a real purpose: continuous optimization of digital marketing tactics and ongoing performance monitoring. Attribution helps you understand which creatives perform better, which audiences engage more, and where your campaigns might be leaking efficiency. This has an even greater value when post-view activity is tracked, mostly possible within the “walled gardens” and it is used to drive campaign optimization. Know its limitations in cross channel measurement, adjust conversion values to align more closely to your business and you will have a proper data feed to fuel your ai driven campaigns in advertising platforms. However, in-platform attribution will never tell you the truth about the cross channel impact of your marketing.

Attribution cannot inform budget allocation, measure incrementality, or establish causality. For those decisions, attribution data is directional at best—a signal, not an answer.

Understanding this distinction is liberating. It frees you from the endless pursuit of “fixing” attribution and allows you to focus on what actually matters: understanding whether your marketing is growing your business.

The companies that will win in the next decade aren’t the ones with the most sophisticated measurement implementations. They’re the ones that understand the limitations of each methodology and have built measurement frameworks and system that capture what truly matters.

Subscribe now

Industry Updates

McKinsey: CMOs and CFOs must unite on marketing ROI

A recent McKinsey report warns that companies may be spending millions on marketing campaigns without understanding their direct business impact. The report calls for CMOs and CFOs to collaborate more closely, using AI-powered analytics to bridge the measurement gap. Without clear ROI frameworks, marketing technology risks becoming “a digital money pit.” Read more

Meta on calibrating MMM with incrementality

Meta has published guidance on how to calibrate Marketing Mix Models using incrementality experiments. MMM provides the strategic view, while incrementality tests provide causal validation. Read more

Video: How to make marketing analytics a true business partner

This Dominic Williamson’s conversation with Recast’s Michael Kaminsky on stopping measurement that kills growth is a great watch. His core argument: measurement systems optimized for precision often sacrifice growth for accuracy, which goes against what marketing measurement is all about.

AI shopping agents break the attribution loop

As AI shopping assistants become more prevalent, traditional attribution is facing an existential challenge. When an AI agent recommends and purchases a product on behalf of a consumer, there’s no “last click” to attribute. Marketers must now ensure their brands are known to AI systems—a fundamentally different challenge than optimizing for clicks. Read more

BCG’s four-legged approach to marketing ROI

BCG has published a framework advocating for multiple measurement approaches: Marketing Mix Modeling, incrementality testing, attribution (used appropriately), and brand tracking. The key insight? No single methodology captures the full picture. Companies that triangulate across methods make better decisions. Read more

System1 and Effie: The Creative Dividend

System1 and Effie have published The Creative Dividend, a new book analyzing 1,265 campaigns representing $139 billion in market share, combined with creative testing data from 200,000+ consumers.

The headline finding: creativity and media together account for 60.1% of campaign business results on average—and up to 98.3% in some categories. Neither works in isolation.

The book introduces Excess Share of Creativity (ESOC): a metric capturing how much creative advantage actually enters the market once media support is considered. The key insight? Profit doesn't rise linearly—it accelerates. As ESOC increases, the likelihood of reporting profit growth increases exponentially.

The implication is counterintuitive but powerful: "good enough" creative is often the most expensive choice. Mediocre creative requires more media spend to achieve the same results. Great creative multiplies the return on every media dollar.

For marketers still fighting budget battles, this provides the evidence: cutting creative quality to save money often destroys more value than it saves. The barrier isn't belief—41% of marketers say creativity is seen as a risk. The barrier is confidence in proving it works.

Download the book

Chart of the week

Winter Olympics happened a few weeks ago in Italy. It’s hard to assess popularity unless we compare to other topics. Below the Google Trends searches of 3 different sports events that happen on 4 year period. Indeed football is a massive sport!

Oldies but Goldies

John Wanamaker’s immortal quote (and its modern interpretation)

This quote, attributed to department store magnate John Wanamaker in the late 1800s, has been used for over a century to justify better measurement. The irony? Attribution was supposed to solve Wanamaker’s problem. It didn’t.

Attribution tells you which clicks preceded conversions. It doesn’t tell you which advertising was wasted. In fact, it often leads to the opposite conclusion—that the advertising which attributes best (lower-funnel, demand capture) is the most valuable, when in reality the advertising that attributes poorly (upper-funnel, demand creation) may be doing the heavy lifting.

Wanamaker’s quote remains relevant not because we lack measurement tools, but because the fundamental challenge of understanding advertising’s true impact requires more than tracking clicks. It requires controlled experiments, statistical modeling, and a willingness to accept uncertainty.

A century and a half later, the marketers who embrace this uncertainty—who build measurement systems designed to learn rather than to prove—are the ones who will finally know which half is working.

Let us know, what are you thinking?

Share Quantified Nation

• Cómo afecta la elasticidad de los precios al marketing

• El artículo sobre Redes Neuronales de Nueva Generación (NNN) para MMM

• El ROI del marketing de influencers

• El Playbook sobre reporting de Google Analytics

• Artículo de Journal of Marketing Research sobre cómo las promociones y la publicidad afectan a la elasticidad de forma contraria

Seeing some recent conversations on the topic (summarized here by Koen Pauwels) there’s a great deal of apparently contradicting claims. Where’s the truth?

Our thesis is that the reach/frequency question has some inherent technical complexities and nuances that make arriving at a “truth” difficult. However, much of the messiness of the conversation and the difficulty to see a clear picture comes from confusion in terms, mixing of concepts…

The right language is the first step towards a common understanding. Let us explain the different aspects of this topic from several perspectives.

Fair warning: this is a longer read than usual, but the topic deserves it!

The definition of terms and its convention

Consumers receive advertising impacts: that’s the object of our study. But of course not all impacts are the same, so we can start grouping them according to different characteristics.

One way of doing that is dividing them into reach vs. frequency impacts. What impact counts as “reach” and what impact counts as “frequency” is based on convention. Of course reach means 1st impact and frequency are subsequent impacts, but the specifics of what that means will depend on:

• The time dimension: reach/frequency can be calculated at the day, week, month or campaign level. The longer the periods, the more the balance tilts towards frequency.

• The same goes for channels. Because it’s so difficult to measure cross-channel, often reach/frequency get reported (or talked about) in isolation or as aggregation of similar channels—it’s not usually a comprehensive cross-channel measure.

Making those explicit and clear is our first recommendation.

What do we mean by “optimal”?

This is the second term where we need to be explicit and precise on the meaning.

Naturally, the question the business asks is “what’s the optimal frequency?” Fair question, but totally incomplete. What do we mean by “optimal”?

• Maximizing the ROI for a given spend?

• Maximizing the topline revenue if we change the spend?

• Having a campaign that is big enough to measure an impact (more or less sophisticated)?

• Optimal for a channel? For a campaign?

All proper conversations should have crystal clarity on what we mean. Most often, that’s not the case.

A good analyst is the one that helps the business ask the right questions. This is a great opportunity.

Surfacing the apparent contradiction

Many apparent contradictions could be solved by a proper definition of these aspects.

What is “optimal” for driving the mass of a business (effectiveness) is totally different from what is “optimal” to drive the ROI of a given spend (efficiency).

We often see reports advocating for higher frequencies because they take a stance of maximizing effectiveness, which is 99% of times the right approach and what business cares about. That’s mostly the case for those advertising channels where media buyers may be too shy on investment: often these channels can still absorb significant amounts of investment at a marginal ROI bigger than the average ROI of cross-channel. In that case, there’s a clear effectiveness argument to be made and a higher frequency is just the consequence of that.

Is that all? Not really, there’s something more profound on these macro recommendations and frequency vs. the consumer reality.

Also, we’ve been told for years by people at the Ehrenberg-Bass Institute that we should be maximizing reach. Why all this talk on frequency now? How does it all square?

A simple consumer model

This topic can get messy really quick because it has so many elements: media planning, media costs, consumer behaviour, measurement challenges…

What we see in macro observations in A/B testing or MMM has to be consistent with the aggregated behaviour of individual consumers. After all, it’s about consumers that this is all about.

A good intellectual approach is to think of single consumers first, as aggregation of consumers should have consistent behaviours unless we are ready to accept “circles upon circles” arguments such as network effects or similar “emergent behaviours”2. Also, simplicity can be a great starting point and we can always add complexity later.

In other words, any insight that we get from analyzing macro data that can’t be explained in simple consumer terms should be taken with suspicion.

Let’s start with a very simple mental model. Let’s say all consumers have equal predisposition towards buying a brand, have similar media consumption patterns… Let’s dream that a media planner could decide who sees the advertisement. Big assumptions, but helpful for the exercise.

Then we deliver our 1st advertising contact. It would not matter who receives it since consumers are homogeneous, so no decision is to be made.

Then our 2nd contact—who should you deliver it to?

This is the fundamental question.

One could have different opinions. From the data we have seen in our careers, data collected doing proper brand lift meta-analysis while controlling all the variables, the optimal is to deliver the message to a new person.

You reader may have doubts on this or have seen other kinds of data, but let’s do a “reductio ad absurdum”: If you are a believer in giving a 2nd contact to the same person—what if it was 5, 10 or 15 impacts, would you do the same? Feels weird, right?

All of us would agree that 100 contacts to 1 person is worse than 1 contact to 100 people.

The problem is on the grey area: 20 contacts to 5 people, 5 contacts to 20 people or 1 contact to 100 people? Any answer different from the last one implies that there is some kind of “tipping point” in consumer decisions.

Is it reasonable to think of tipping points? In other words: is there a number of contacts where consumers don’t have any advertising impact and suddenly there’s a threshold where they will certainly buy our product?

Of course not: what every advertising impact does is to increase the probabilities of a purchase. There’s no tipping point, but rather an accumulation of higher probability as we increment the frequency.

Consequently: it’s clear the 1st impact is the best one, and an ideal scenario would maximize the number of “1st time impacts” vs. any other option.

Limitations of this model

As in all model representations, they hold vs. reality as long as some conditions are met. Several phenomena could break our assumptions:

• Consumers have different predispositions to consume certain brands. This makes reaching some people more valuable than reaching others. This is due to fluffy “segmentation-affinity” effects, but there are also hard elements, like heterogeneous distribution of a product.

• People have different media consumption patterns. This introduces a cost efficiency element into the planning.

• Advertising delivery has frictions: one does not go 1:1 with every single consumer but often impacts several at the same time (think TV, outdoor…), there’s never perfect information on who is behind the screen, friction between the planning and the execution of a campaign…

All these elements are true and that’s what justifies the need for sophistication in planning. But they can’t be above the fundamental truth that the 1st impact is the best.

More is more, but how much more?

But we do keep on seeing charts that say that higher frequencies drive more business. How come?

Of course a frequency of 3 produces bigger results than a frequency of 1. But our role in optimization is: what’s the better alternative?

YouTube’s BrandLift database is probably the best in the world for these matters: massive sample per experiment, proper test-control designs of the cells, a limited number but constant set of KPIs, thousands and thousands of cases… and most importantly: metadata on execution levels.

Google published data from that database in 2017, which can still be found here or in this presentation. We have no reason to think that these insights have changed significantly.

What do we see in the chart? The most important numbers are those 1.5 to 1.8 indexes for the 2nd impact: it’s always less than 2x. The first impact is the best, the second delivers always less. Same goes with the subsequent ones.

This extremely high quality data proves exactly what we discussed in the previous point.

(See minute 16 of this presentation)

An apparent paradox

How can the F=1 be the best impact and many of the papers going out today say that it’s better to have F=3 or higher?

This is a consequence of media planning. It does not mean you should be optimizing for F=3 at the expense of reach!

In reality, you can’t decide to build exclusively reach or frequency. Reach and frequency are being built simultaneously, in a stronger or weaker way depending on your media planning capabilities (frequency capping, smart scheduling of TV insertions, smart combination of different platforms…).

Saying that a business is better off by having a frequency of 3 in an underinvested channel with great ROI (vs. putting the money in other channels with lower ROI) is absolutely right and true. It’s just that the point is that the underinvested channel deserves more, and if higher frequency is the consequence, so be it.

On the measurement precision limitations

Do not let the technical limitations of aggregated models (such as MMM) be inconsistent with real consumer behaviour.

From what we said before, it follows that an S-Shaped curve ultimately contradicts the consumer behaviour model we described earlier. But not only: it also contradicts the BrandLift data we have.

Simple explanations are often the right ones. Let’s not be forced to increase the complexity of the consumer model if the simplest explanation is that aggregated models such as MMM struggle to properly measure low media execution levels.

Most often the S-curve is explained as “first you don’t see a lot of impact because there’s not enough consumers impacted by the investment, then you see it ramp up, eventually there’s diminishing returns.” Easy to understand but not true to reality: the earlier stages of the media investment do deliver value too—actually it’s likely that they deliver the most value. It’s just that we can’t properly measure because of the low signal-to-noise ratio.

The S-curve is a reflection of our limitations of measurement.

Reflecting these into modeling

Most often, what goes into the models is a Hill function, be it on total impacts (traditionally) or on some subset of the impacts (such as in Meridian, which has linear assumptions on reach and Hill function for frequency).

This is how a Hill function looks like. Note there are two parameters (K and S in this case). Here we have created a simple Gemini Canvas interactive tool to play around.

For the sake of this topic, we care about the S parameter. Any value >1 would mean some degree of S-Curve behaviour, while <1 would get us a classical diminishing returns curve.

It is our view that values <1 are the only ones that are consistent with true human behaviour.

(From the Google Meridian R&F paper)

A proposition

So, how do we move forward in practice?

Modern media measurement is all about combining different data sources and methods to give the best answer. You all know about calibration of MMM using A/B test data.

Following this approach: what if we incorporated brand-lift results of frequency by level as priors to the Meridian model for the K and S parameters?

Seems like a good way forward.

Subscribe now

Industry Updates

Mark Ritson end of year reflections

Resist the push to make predictions, better approach to reflect on the current events. #1 is about pricing, have a look at QN19 where we talk about it.

IPA Effectiveness Awards 2026 opens with a new paradigm

The IPA Effectiveness Awards 2026 are now open for entries, with a notable shift in focus. As Charlie Ebdy from Omnicom Media Group UK notes, “the advertising industry’s shift toward digital media has outpaced its ability to prove effectiveness.” The Awards are evolving to meet this challenge, with early registration closing January 9th. This year’s theme asks a fundamental question: do digital ads really work? A timely reminder that effectiveness proof remains the industry’s biggest challenge. More details here.

Google Meridian adds pricing and promotion variables

Google’s open-source MMM framework Meridian received a significant update in late 2025, now including non-media variables like pricing and promotions, channel-level contribution priors, and enhanced binomial adstock decay. For those of us working with reach and frequency modeling, the enhanced priors capability is particularly relevant—it opens the door to the kind of calibration we discussed in this edition. Read more.

Ad Age: Metrics that will matter most in 2026

Ad Age’s annual forecast highlights incrementality and attention as the metrics to watch this year. The report notes a shift away from traditional impression-based measurement toward proving actual business impact—a theme that resonates deeply with this edition’s discussion on frequency. As the industry matures, expect more scrutiny on whether those extra impressions actually drive incremental value. Full article.

Chart of the week

Christmas is over and surely you have been tempted with some sweet delicacies. There’s one Italian export that has been increasingly popular on all shelves.

Have we reached “peak Panettone” yet? Apparently not: yet another year with record interest in this product.

View on Google Trends

Oldies but goldies

Erwin Ephron and Recency Planning

Before his passing in 2014, Erwin Ephron was perhaps the most influential media planning thinker of the modern era. His 1995 concept of “Recency Planning” fundamentally challenged the industry’s obsession with frequency thresholds.

The conventional wisdom of the time—rooted in Herbert Krugman’s “three-hit theory” from the 1970s—held that consumers needed multiple exposures before advertising could work. Media plans were optimized for “effective frequency,” often targeting 3+ exposures per person.

Ephron turned this on its head. His argument was elegantly simple: advertising works by reminding people who are already in the market. Since you never know when a consumer will be ready to buy, the best strategy is to be there continuously—to maximize the probability of reaching someone at the right moment. Reach, not frequency, should be the priority.

As Ephron put it: “One exposure is not a waste. It’s all you need to remind a person who is ready to hear.”

His 1997 paper “Recency Planning“ remains essential reading. It anticipated by decades the debates we’re still having today about the diminishing returns of frequency and the primacy of the first impression.

The irony? Thirty years later, with infinitely more data and computing power, we’re still fighting the same battle Ephron won on paper. Perhaps it’s time to re-read the original.

Share Quantified Nation

Indeed one of the greatest advantages of having a strong brand is the ability to charge higher prices and see how your demand gets little impacted, or being more resilient to price changes.

However, this topic can get technical, and it’s easy to confuse the terms. Remember that price elasticity already measures sensitivity to price variations. So, when we say “good marketing decreases price elasticity,” we are technically talking about a variation of a variation.

Why does this matter? To clear things up, let’s look at a hypothetical scenario. The actual numbers don’t count; it’s just for the sake of the argument.

• A product that sells currently 100 units for 10€

• 3 scenarios of elasticity: Price elasticity of -0.8 (inelastic) and -1.2 (elastic) and -2.5 (very elastic)

• COGS 2€ per unit on “high margin scenario” vs. 8€ per unit on “low margin scenario”

With those datapoints, how do revenue and profit change if we apply the usual pricing elasticity formulas?

Let’s start with the revenue. In this case there are only 3 possibilities as the margin is irrelevant. The chart is absolutely predictable and determined by the elastic-inelastic starting situation. Elastic brands decrease revenue if they price up, inelastic brands increase revenue. That’s it.

When we look at profit the situation gets more nuanced.

• We see that for low profit margin (blue) actually the impact that the different price elasticity could have is minimal.

• On high margin scenarios (green) the different price elasticities can shape the outcome. Around an elasticity of -1 there's more to gain or lose, it's a "flatish" line. That's true for elasticities around abs(-1.2 to -1.5). If elasciticy is higher than abs(-2), then increasing prices at high margins is not a good idea.

What does this mean in practical terms?

For those scenarios with low margins, raising prices is a critical decision and will deliver profit increases even if at the cost of revenue. Just consider raising prices to get out of the profitability trap and worry less about elasticity. In this situation, the literature on how marketing changes elasticity is secondary.

But that’s only the beginning. You should wonder… “why am I in that bad situation of low margins?”. Kantar BluePrint for Growth would say “because you did not have enough pricing power”. In a way, by not having invested into a strong brand you got yourself into that profitability problem. Marketing alone and its power on shaping the elasticity curve won’t solve your short term issues, just take the hit and increase prices. They key is that you reinvest those extra profit into brand-building activities to strenghen your position in the future.

For those scenarios with high margins, then the relationship of marketing with price elasticity are of upmost importance. This comes to the details and each company should do the maths themselves.

Generally speaking, if the elasticity is low (inelastic of low values of elastic) there is more freedom to play along the pricing curve to become a more mass or more niche brand, while keeping profit more or less similar.

If your pricing elasticity is very high and your margin is high too, since you are already making a lot of money on every product sold, increasing prices won't pay off in terms of profit (the revenue loss will dominate). In that case, the most sustainable decision would be to invest in marketing in order to decrease your price elasticity and then increase prices.

Side note: price elasticity is as complex metric from a mathematical standpoint, it’s not as simple as it looks like from the conceptual POV. For example, it should not be confused with the derivative of the revenue vs. pricing graph! For the calculations of this newsletter we used the "midpoint elasticity" formulas. Other definitions like the “Point elasticity” are even tricker in math treatment, as they are not even symmetrical.

Subscribe now

Industry Updates

NNN: Next-Generation Neural Networks for Marketing Measurement

A radical new approach on methodologies, absolutely worth reading even if it will take time to materialize, if it ever does. Fascinating because it applies some of the techniques that enabled the GenAI revolution into the MMM world.

In particular:

• Attention (as understood in the LLM world, not in the advertising attention world) as a substitute of fixed adstock decay rates

• Unlike traditional MMMs that only use scalar inputs (e.g., $100k TV spend), NNN uses embeddings to represent qualitative data including ad creatives or search queries, both to distinguish “good” and “bad” impressions or investment

• The model is “nested” because it explicitly learns the causal chain from marketing investment to demand estimated by search queries to finally sales

Of course all comes with some concerns, but those do not undermine the value of this paper and how we should reflect about it. In terms of concerns these could be:

• Overkilling with top-notch models the already very constrained datasets that source MMM (the 3 years x 52 weeks points). Overfitting?

• Loss of interpretability. MMM inputs like impressions/GRPs/spend are not only practical because it’s the data that it’s available, it’s also practical because it’s they way that marketers can act upon the results.

• Creative is an input or an output in a model? Both could be argued.

If we were to compare it with Meridian, NNN would be an “experimental” approach, it requires large amounts of data to work effectively and is a “black box”, harder to explain why it made a prediction. Potentially, it would be a good solution for savvy brands with large amounts of unstructured data like ad creative visuals and search queries data.

IPA study on Influencer ROI

The IPA has published a meta-analysis of influencer campaign results based on MMM across different countries. The full report can be read here. Note that even though the IPA is based in the UK, this analysis includes a mix of 220 cases across a variety of countries.

• In general terms, the channel overall behaves very close to the average, with better results on Telco and SkinCare categories.

• When long-term effects are included, the channel achieves a 151 ROI index vs. the average, which is equivalent to a 3x long term multiplier. However, it’s important to note that this is based on a subset of 18 observations out of the 220 of the database, because measuring long-term is something still unusual in MMM projects.

• Give the nature of the channel, there is some unpredictability. Media optimisation parameters don’t clearly predict results. On the other side, there is a small % of campaigns that delivern an exceptional ROI, meaning that potencially there’s a jackpot to be won if the right influencer, topic, execution, virality… gets into place.

Google Analytics Reporting Playbook

Without any doubt, the previous version of Google Analytics (Universal) gave users confidence about their data. Consent modeling (or modeling in general) did not exist, sampling was “simple”, reports were predefined, BigQuery export had very similar data to the interface or the API and it was a continuation of the earlier Google Analytics versions (some folks might remember remember the “Classic” version).

This is not the case today. Modeling is required today to understand user behavior, aggregated data is a very valuable asset which is not exported to BigQuery, behavioral and advertising data requires separation, explore reports give the granularity but standard reports use preloaded tables which can be customized for ongoing needs. The list of reporting differences between Universal and GA4 is long and has been evolving over time. While you can stay complaining that GA4 does not work like Universal, it is a much smarter approach to embrace and learn how reporting works today in Google Analytics. The measurement world is much complex it used to be and reporting in Google Analytics is a reflection of such evolution.

The Google Analytics product team has launched this resource which requires reading, study, practice and circle back to ensure you make the most of it. The Google Analytics Reporting Playbook provides a comprehensive overview of all reporting surfaces in Google Analytics.

Reports, Explore, Advertising, Data API and BigQuery export have unique features that require understanding their distinct features to know where to look for insights, whether you are analyzing user behavior across your website or app, or measuring the performance of a marketing channel.

This playbook includes detailed sections on each of these areas, as well as dedicated information on Audiences, Case Studies, Common Questions, and Reporting Gotchas.

This is already an essential resource for everyone relying on Google Analytics to understand their business.

Download, print and take it with you. It’s really worth the time.

Incrementality and Retail Media: IAB guidelines

Retail Media is one of the big winners in 2025 in marketing spend growth, outpacing general digital ad growth, and becoming a core channel as brands shift budgets for its high-ROI, first-party data and availability of conversion reports.

The IAB has published its Guidelines for Incremental Measurement in Commerce Media which makes an effort to provide a common language for the industry’s most confusing metric. This report is essential because it legitimizes a pragmatic approach to measurement, moving beyond the obsession with “perfect” data.

The core conclusion is “Fitness for Purpose.” The guidelines explicitly state that you don’t need academic-level rigor for every single decision:

• For high-stakes strategy: Use strong causal methods (like Geo holdouts or MMM) when allocating budgets or proving ROI to the C-suite.

• For daily optimization: Lighter proxies are acceptable. When tuning a live campaign, speed often outweighs precision.

Crucially, the report draws a hard line: ROAS is an efficiency metric, not a proof of impact. If you are tired of circular debates where “Attribution” is confused with “Incrementality,” this document is a great resource you should point to.

Chart of the week

New Year is coming and with it, habit change resolutions. How well do they fare as the months pass by? I think we all know the truth, and this Google Trends chart shows it:

Oldies but goldies

The Long-Term Impact of Promotion and Advertising on Consumer Brand Choice

Since we are discussing how marketing shapes price elasticity, we have to look back at this classic from the Journal of Marketing Research, published in 1997. Mela, Gupta, and Lehmann analyzed over 8 years of data to answer a critical question: Do our marketing actions change how consumers react to price over the long run?

Their findings are the scientific backbone of today’s newsletter:

• Advertising reduces price sensitivity: Consistent advertising helps brands differentiate themselves, making consumers less sensitive to price changes (inelastic).

• Promotions increase price sensitivity: Reliance on price cuts trains consumers to wait for the “deal,” making them more sensitive to price (elastic).

It’s a powerful reminder that while a price cut might win the quarter, it will slowly break your brand pricing power forever.

Share Quantified Nation

• La importancia del análisis descriptivo para entender las grandes dinámicas del marketing y su impacto en el negocio, como paso previo a la modelización

• Updates de industria: nuevos informes y premios Effie

• El artículo de Booking sobre selección de variables para el análisis de inferencia causal

• El nuevo informe de Ehrenberg-Bass sobre el “propósito de marca”

• El estudio de PyMC labs sobre Meridian vs PyMC

• Chart of the week: incrementalidad

This is not always the best solution: working on descriptive data could make the point. The reality is that sometimes, less is more. While sophisticated models have their place, we risk overlooking the power of simple, descriptive analysis. Consider this: instead of spending weeks refining a complex model based on three years of data, what if we broadened our scope? By gathering data spanning five or even ten years, we can paint a richer, more comprehensive picture of long-term trends. A well-executed descriptive analysis, focusing on clear visualizations and straightforward metrics, can reveal patterns that are often obscured by the complexity of advanced models.1

Simple descriptive charts are easy to communicate and tell good stories. Descriptive analyses are incredibly effective in communicating insights to stakeholders. A clear, well-structured chart can be far more accessible and impactful than a complex model output. These visualizations tell compelling stories, enabling us to convey the impact of media strategies in a way that resonates with both technical and non-technical audiences. This storytelling capability is essential for gaining buy-in and driving strategic decisions.

A criticism to this approach could be saying that just by looking at the data we will only find correlation and not causality. However, this concern can be mitigated by our deep understanding of the business landscape. We, as experts, can contextualize the data, explain the underlying mechanisms, and provide compelling narratives that makes the “causality debate” irrelevant. By combining our analytical skills with our business acumen, we can confidently address these criticisms and validate our findings.

The Descriptive First Approach

Before a single regression is run, the best analysts build a simple, descriptive view of how the business has evolved in the area of analysis. When doing a Marketing Model Mix, this means having a full understanding of the media execution and of the evolution of the KPI to model along with evolution of the business and related KPIs. This is the foundation of a successful measurement project that closes the application gap.

Grounding an MMM in Reality: Too many Marketing Mix Models begin with data wrangling and end in a black box. A better process starts with a killer chart. Imagine plotting three years of your total sales, market share, and media spend on a single timeline. This simple view reveals the macro trends that a model focused on short-term variance might miss. It forces a conversation about the business context before the modeling begins, which is essential for avoiding the “insulting recommendations” that are statistically sound but practically useless.

Therefore, as much effort should be put into the right models as in collecting long-term datasets that tell great stories. This is especially true for the long-term effects of media. In conclusion, while advanced modeling remains a valuable tool in our arsenal, we must not neglect the power of descriptive analysis.

The biggest risk in descriptive analysis isn’t the data, it’s the analyst. A compelling chart is a powerful tool, but it can be used to support any narrative. The analyst must choose their role: are they a storyteller, honestly trying to find the clearest narrative the data supports? Or are they a lawyer, cherry-picking data to build a case for a pre-conceived conclusion? We’ve all seen the “7 deadly sins of data-driven marketing” and this is where Pride (using only the data that makes us feel good) and Envy (biasing data to get the results we want) come into play.

An example would be looking for the impact of media into brand metrics. You know the tools: nested models, baseline modeling… That’s the instinct of a seasoned analyst. However, anyone that has seen brand tracking metrics quickly realizes that the variability of this data when looking at 1-2 years is often low and does not tell a great story. Any good modeler should first plot the descriptive data of those brand metrics to see what’s the deal, and very often they are not that exciting. Why spend so my time to find the needle in the haystack, torturing the data to find a subtle pattern? Instead, look at the long term: 5 years is better than 3 years, and why not 10 years? That kind of data will always tell a more interesting story on the brand metrics. And then just overlay the marketing investment across the years: no model, just the metric. If you are lucky you may have found a killer chart just using descriptive data, which will impact 1000 more than a complex nested model looking at 2-3 years of brand metric performance.

The Future is a Human-AI Partnership. Looking at the LLM’s current ability to create code, a next frontier is easily taken with any kind of data analysis, descriptive to begin with. Tools like Gemini, Claude or ChatGPT can act as incredible brainstorming partners, helping an analyst explore hypotheses at lightning speed as well as developing the underlying code to analyze the data. Specialized tools like Julius AI can create visualizations from plain English commands, freeing up the analyst to focus on interpretation, not coding.

AI will not replace the analyst. It will only make the analyst’s strategic judgment more important. The AI can generate a hundred charts, but only a human with deep business context can identify which one is the killer chart—the one that tells the true story and drives the right action. The data point which makes the CFO want to know more and finally have this conversation about budget increases.

What this means, Marketing Leaders?

• Hire for Business Acumen: When you build your team, look for analysts who are deeply curious about the business. Technical skills can be taught; strategic thinking and storytelling ability are much harder to find.

• Embrace the Co-Pilot Model: Encourage your team to work with AI tools to accelerate their workflow. The goal is to free them from tedious tasks so they can spend more time on what truly matters and provide better guidance for business decisions.

The future of measurement isn’t about abandoning or leaning more into complex models. It’s about keeping a balance. The hunt for the killer chart, grounded in simple, powerful descriptive analysis, is a strong step in this direction. It grounds our work, sharpens our hypotheses, and makes all our subsequent, more complex analyses truly effective.

Find the chart, the combination of metrics or metric growth that will make your audience raise their eyebrows while explaining something obvious

Subscribe now

Industry Updates

“The Future of Measurement” Conference by ANA

The USA association of advertisers is organizing a physical/virtual conference about the futre of measurement on September 8-10. You can find more details in this link.

Marketing Week’s “Language of Effectiveness”

The 2025 edition of this research on how advertisers measure marketing is out. In this link there are several articles discussing the key insights of this research partnership of Kantar and Google.

Spanish Effies

The 2025 Effie winners in Spain (jury composition) were announcedd October 23rd in a gala in the the Royal Theather of Madrid. The Grand Prix was awarded to Pikolin, the prize for creative agency was for DDB&14 Group and the prize for media agency was for Link to the full list of winners

Covariate Selection in Causal Inference

When running A/B tests isn’t an option, we have to rely on observational data. But what variables should you control for to estimate a causal effect? This excellent article from Booking.com’s data science team provides a practical guide on “good vs. bad controls.”

It explains how controlling for the wrong variable—like a mediator (a variable on the causal path) or a collider (a common effect)—can introduce significant bias, sometimes even more than omitting a variable. The post offers clear takeaways (e.g., confounding bias is usually worse than collider bias) and rules of thumb for selecting covariates to get more accurate causal estimates from your observational data.

Who cares about Brand Purpose?

A new report from the Ehrenberg-Bass Institute questions the effectiveness of “brand purpose” as a marketing strategy, finding that consumers are mostly unaware of these initiatives.

The study surveyed almost 3,000 consumers in the US, UK, and Australia about 14 “exemplar” purpose-led brands.

• Awareness is low: On average, only 18% of consumers could correctly link a brand to its stated purpose.

• Guessing is high: The researchers included “dummy” purposes and found an average guessing rate of 9%. This suggests genuine awareness is likely less than 10% for most brands.

Key Takeaway: The report concludes that building awareness for a brand’s purpose is incredibly difficult. It cautions marketers to have realistic expectations, suggesting that if growth is the goal, investing in proven drivers like Mental and Physical Availability is a more reliable strategy.

PyMC Labs compares PyMC-Marketing and Meridian (guess the result?)

For brands choosing an open-source Marketing Mix Modeling (MMM) framework, PyMC Labs has published a technical benchmark comparing their own PyMC-Marketing against Google’s Meridian.

The comparison used synthetic datasets mimicking four scales, from a small “Startup” to a massive “Enterprise” (50+ geos, 30+ channels).

• The benchmark concludes PyMC-Marketing is the “more robust choice,” showing advantages in speed, accuracy, and scalability.

• Performance: PyMC-Marketing was 2-20x faster and cut the error rate on channel contributions in half compared to Meridian.

• Scalability: While both libraries handled smaller datasets, only PyMC-Marketing could successfully run the massive “Enterprise” level dataset; Meridian failed to converge.

• Model Fit: PyMC-Marketing consistently produced a better model fit (higher R², lower MAPE) and avoided the systematic errors (autocorrelation) seen in Meridian’s results.

The only area where Meridian had an edge according to this benchmark was in creating smaller model files on disk, though this comes at the cost of sacrificing accuracy and speed.

While the comparison is purely quantitative, the selection of the features in this benchmark will not provide a definitive answer to teams trying to decide. Meridian includes additional features which make it a better fit in many cases not considered in the benchmark.

Meridian excels when using geographical data to add granularity, employing a hierarchical model for a better prediction. Additionally, adding search query volume, reach and frequency metrics provides better modeling capabilities, which were not fully considered in the benchmark. The solutions also differ in their approach to media effect parametrization, seasonality and while both are bayesian approaches, the inclusion of the priors follow a different approach.

Using these features wisely can also give you a higher edge with Meridian so ensure you follow a wholistic approach to decide which framework to use.

Chart of the week

Incrementality is the latest new word in the trade (link). The average search volume has roughly doubled in the last year with some really big punctual spikes driven.

However, just as MMM is fashionable now but not really new, it’s the same for incrementality! Ben Dutter in LinkedIn made a great reflection on why it’s really paradoxical that after so many years and advances in marketing measurement, we ended up talking about incrementality while it should always have been the core of it. In the end, there have been so many fooled by ROI/ROAS attributed numbers that reading incrementality metrics often requires some preparation.

Oldies but goldies

While this video is pretty recent, it is a great practical summary on what we have been learning about proper brand building over the decades. Mark Ritson’s manners are shocking and make the delivery of his presentations really engaging… But his views are nuanced and not dogmatic. This video summarizes how brands should be built, integrating insights from the different schols of thought.

Share Quantified Nation

From a “searching for the truth” standpoint, they make perfect sense. However, for practical business situations that demand decisive action, do the three vertices of the triangle carry the same weight?

They don’t. For most businesses, the triangle is fundamentally unequal. One side is longer, heavier, and far more critical than the others.

This is not to challenge the view that all three elements have a role to play, or that a well-designed experiment is the closest we can get to a perfect measure of “true incremental impact.” Our argument is based on a pragmatic reality: in terms of driving transformative business strategy, MMM is likely the most useful and impactful tool for companies.

The Foundation: Why Modeling and MMM Carries the Most Weight

While experiments provide causal truth on a small scale and attribution offers a high-frequency pulse on digital tactics, MMM is the only framework that provides a panoramic view of the entire business.

• It’s Strategically Holistic. An MMM is a must-have for any organization spending regularly on more than two marketing channels for the last couple of years. Its purpose is to answer the C-suite's most pressing questions: "What is our optimal total budget?" and "How do we allocate it across our portfolio to maximize growth?". If your MMM is not serving as a strategic guide to budgeting and you spend regularly on more than two channels, you have a massive optimization journey ahead of you.

• It’s Flexibly Comprehensive. Unlike the rigid structures of other methods, MMMs can be adapted to every business situation. The choice of variables, the modeling of brand effects, the inclusion of competitor actions, pricing, and economic factors—this flexibility allows it to mirror your business's unique reality.

• It Fosters the Best Business Conversations. Because an MMM incorporates all business drivers, it forces a conversation that transcends the marketing department. It becomes a shared language between Marketing, Finance, and Operations, connecting marketing spend not just to sales, but to profit, market share, and long-term value.

Experiments, by contrast, are often related to very specific questions and are unlikely to deliver a transformational impact on their own. The same goes for attribution, which, by design, delivers optimizations within digital channels. As important as this is, it is not an analysis of the fundamental drivers of the business. Companies using post click attribution as their main strategic input are at least ignoring 75% of the real impact of marketing and the marketing channels they could be using to grow their business.

MMM aims to incorporate the business knowledge needed so that decision makers do not have to lean that much on their own, valuable but limited historical experience or biases and can incorporate those in the model so that in the end they can make better business decisions.

Deconstructing the Other Vertices

If MMM forms the strategic base, incrementality and attribution are the supporting—but unequal—sides. Understanding their distinct roles and limitations is crucial.

Incrementality: The Costly Pursuit of Causality

Incrementality setups are the undisputed gold standard for understanding causality. They answer the question: "Did my marketing action cause this business outcome?". However, this truth comes at a price. These experiments are costly, both in the technical expertise required to get them right and in the significant opportunity costs associated with having holdback groups that are not exposed to your marketing.

Geo-experiments are often hailed as a cross-platform solution, but best practices for their adoption are rare to find, and they are expensive and complex to implement correctly. Despite the cost, having a robust understanding of causality is non-negotiable when justifying significant shifts in marketing investment.

Attribution: A Misleading and Misunderstood Tool

"Attribution" is perhaps the most misleading vertex of the triangle. Do we mean a cross-channel Multi-Touch Attribution (MTA) model or the attribution reported within a specific advertising platform? The distinction is critical.

• Multi-Touch Attribution (MTA) is often argued to be increasingly limited. In a world with less identifiers, its ability to stitch together a comprehensive user journey across platforms and devices is severely reduced to a limited post-click view of the world. The days of id-based attribution are gone and those trying to stitch marketing actions and business outcomes with personal ids cannot be for the purpose of understanding the value of these marketing actions. However, modeling has come into the rescue to multi-touch attribution and these models are embracing new methodologies to regain relevancy. Nevertheless, their scope remains narrow.

• Platform Attribution is even more problematic when used for comparison. Trying to compare the ROAS from Meta with the ROAS from Google is an exercise in futility—it’s not apples and oranges; it’s apples and pears. Each platform operates within its own walled garden, measuring its logged-in users with its own proprietary methodology. These numbers are fundamentally different and impossible to compare directly. Their true purpose is to provide a consistent signal to feed their internal bidding algorithms, not to inform your cross-channel budget allocation.

Platform Incrementality

Similar thoughts come to mind with platform-reported incrementality lift studies, where ever-changing, black-box methodologies make cross-platform comparisons unreliable. These are, in turn, extremely valuable tools when you want to have a good grasp of what is the incremental impact of a campaign on different KPIs, with different setups or creatives. A must have in your toolbox which has to be taken with rigour and across different moments in time to have a consistent idea of incrementality using these tools.

So, What to Do in This Unequal Triangle?

The goal is not to discard the triangle but to use it wisely, acknowledging its inherent imbalance. This means establishing a clear operational hierarchy.

1. Lead with MMM for Strategic Decisions. Your MMM should be the primary tool for annual and quarterly budget setting. It provides the strategic "why" and "how much" for your marketing plan.

2. Use Incrementality to Validate and De-Risk. Deploy costly experiments surgically. Use them to validate the major directional shifts suggested by your MMM. If your model suggests cutting your largest channel's budget by 30%, a geo-experiment is the perfect tool to de-risk that decision before you commit.

3. Combine Them Pragmatically. The textbook approach to combining incrementality and MMM involves using experiment results as Bayesian priors. This is powerful but not the only option. A more pragmatic approach can work: if an experiment shows a channel has a much higher incremental ROAS than its attributed ROAS, you can work to manually adjust or re-attribute conversions from "Direct" or "Organic" channels in your reporting to better reflect this reality. It's a "naive" but often effective way to bridge the gap between models.

4. Leverage Attribution for Tactical Optimization. Accept platform attribution for what it is: a valuable, but siloed, tool for in-flight campaign optimization. Use it to improve performance within a channel whose budget has already been set by the MMM. Do not use it to decide whether to move money from one platform to another.

Beyond the Triangle: The Blind Spots of a Short-Term Focus

Perhaps the biggest danger of the measurement triangle is that even a perfectly executed framework can create strategic blindness. All three methods are inherently focused on measuring relatively short-term effects. By focusing only on this, you risk ignoring the fundamental laws of advertising that drive long-term, sustainable growth.

The triangle is often silent on:

• The Power of Creativity: As we've discussed before, creative quality is a massive driver of incremental sales. A framework that only optimizes media spend without accounting for the quality of the message is missing the biggest piece of the puzzle.

• The Value of Brand Building: The triangle is poor at capturing the slow, cumulative impact of brand awareness and mental availability. A relentless focus on short-term, attributable conversions can lead companies to underinvest in the brand-building activities that reduce price sensitivity and increase baseline sales over time. An MMM which is not accounting for long term impact through intermediate metrics is also limiting the view at that level and overestimating the short term impact of marketing.

• The Necessity of Reach: The goal of reaching a relevant portion of your total target audience is a cornerstone of marketing effectiveness, championed by the likes of the Ehrenberg-Bass Institute. An obsession with narrowly targeted, high-ROAS segments can starve a brand of the broad reach it needs to grow.

Your Measurement Framework Is Your Competitive Advantage

In this complex landscape, it is tempting to wait for a single, perfect solution that you can simply implement to solve all your measurement problems. That solution does not exist today.

Your competitive advantage today lies in your ability to build and leverage a proper measurement framework that works for your organization and the way it makes decisions. This means having the ability to understand the impact of your main marketing activities, how they interact, and which levers to pull depending on the direction the business is heading. It means planning your budgets with a data-driven approach and optimizing campaigns for maximum impact.

This is not easy. It requires a concerted effort to put together the systems, processes, and—most importantly—the people to thrive in this complexity. The goal is not to find a single source of truth, but to master the art and science of combining these diverse, unequal inputs into a coherent strategy that drives sustainable growth. That is the game we are in.

Subscribe now

Industry Updates

Google Trends API in Alpha

Google has launched an API for pulling Google Trends data, which is one of the most exciting dasets of consumer behaviour in the world. So far the data could only be accessed via the Google Trends tool and in some very selected collaborations. With this new API, access is broadening and it signals that likely it will be even easier in the future. However, it’s still an Alpha and we are unsure on how big it is. In the meantime, specially for modeling-related purposes, remember that Meridian alredy incorporates the ability to include Google searches. Link with more details.

Spanish Effies

It’s summer and this means that the 2025 Effie Jury in Spain (composition) is really busy! A total of 265 cases made it to this stage (long list details). We wish the jury a lot of energy and wisdom!

Why most brands fail at incrementality testing

Recast, a solution for MMMs has a very interesting YouTube channel where they are interviewing Andrew Covato on the topic of incrementality testing. Risk aversion and being afraid to fail stands out as one of the main reasons why incrementality testing is hard to adopt. Moving from attribution based decisions towards incrementality based decisions is hard and risky, easiest move is to stay.

Meta reports on incrementality

Last May, Meta published a new paper called “Building a suite of truth: A hybrid approach to measuring incrementality”. Another spin in the triangle with a broader focus on the experiments piece:

It includes many healthy recommendations on how you should approach measurement which are very well aligned with the ideas in this newsletter. The idea of having calibration and regular experimentation as core competences in your marketing team is a strong callout as opposite as having people making decisions based on experience and gut feeling.

In a more practical direction, Haus published “The Meta Report: Lessons from 640 Haus Incrementality Experiments”, showing strong ideas about the ability of the Meta Ads platform to deliver positive results on incrementality tests by having a full funnel strategy rather than focusin only on lower-funnel campaigns. The idea of the uniqueness of experiments and that testing is unique for each business is also included in the conclusions. This also speaks to the fact that experiment data is messy and that more automation does not mean more incrementality. Food for thought for planning your next campaign to deliver incremental results rather than just attributed conversions.

Google Analytics MCP server launch

The MCP server allows you to connect an LLM like Gemini to our Google Analytics property and talk to our data.

Take a look at it yourself and look at a glimpse of what it is possible with this MCP server in the video below:

Oldies but goldies

We have seen in LinkedIn a book recommendation where many reputable individuals on marketing measurement have left their reaction so we thought this could be the best Oldie but goldies for this summer.

The book is "Reality in Advertising", written by Rosser Reeves in 1961. In a world saturated with complex digital metrics, Reeves' rigorous, evidence-based approach is a powerful reminder that the fundamental challenges of marketing measurement have been with us for decades.

The use of the term "advertising laws", a concept that echoes the scientific, evidence-based principles later popularized by Byron Sharp in his 2010 book, "How Brands Grow". While Sharp's work is grounded in decades of research on buyer behavior, Reeves took a different path, focusing on what makes advertising itself effective. Many of his conclusions, drawn from analyzing packaged goods sixty-four years ago, are quite well aligned with modern effectiveness thinking:

• Ad Penetration (Reach) Drives Performance: Reeves argued that the reach of an advertisement is a primary driver of its success. This is a direct precursor to the modern emphasis on maximizing broad reach to all category buyers, a cornerstone of the Ehrenberg-Bass school of thought.

• Maximize Audience Size, Optimize Frequency: One of his key pieces of advice was to "Reach your audience less often…and make your audience as large as you can". This is a strikingly sophisticated take on media planning, anticipating the contemporary debate around effective frequency and the diminishing returns of bombarding a small audience segment.

Reeves advocated for a simple yet powerful measurement: compare the percentage of people who buy your product and can recall your advertising against the percentage who buy but are not familiar with the ad. The difference, which he termed "pull-over," serves as a direct measure of the ad's effectiveness.

Reeves was relentlessly focused on the link between ad exposure and actual purchase behavior and one of the main pilars for such link was the Unique Selling Proposition (USP) a concept on differentiation that became popular but is challenged today as in most mature markets, achieving a truly unique, sustainable product or service feature is extremely rare. Instead we should prioritize distinctiveness, where what matters more is being instantly recognizable and memorable through consistent and unique brand assets (logos, colors, slogans, jingles, etc.). The goal is not to be different, but to be thought of easily and often in a buying situation.

We would love to hear your thoughts on this. Probably it depends based on your industry and market maturity. While we try to be different, if you choose to read us for distinctiveness, that’s also fine for us. Just leave a comment.

Share Quantified Nation

Navigating this environment isn't about blindly adopting every new technology or methodology, but about developing the strategic judgment to discern between hype and genuine, lasting trends. In this edition, we cut through the noise to analyze the underlying changes that are truly shaping the future of marketing measurement.

If you see your company far from the examples provided, the best you can do is to keep reading Quantified Nation and bring it to your whole team:

1. Increasing demands to do more with less. The C-suite's focus has shifted from "Is marketing delivering a positive ROI?" to "Is marketing delivering the maximum possible ROI for this budget level?". This requires a deeper, more nuanced understanding of marginal returns.

   • Example: The CFO is no longer satisfied with a report showing a 3 campaign ROAS. She now asks, "Your MMM should show us our position on the response curve for this channel. Could we have achieved 90% of these sales with 70% of the budget and re-invested the savings in a channel with higher marginal returns?".

2. Effectiveness will continue growing in importance. The conversation is maturing from short-term attribution to a holistic, business-oriented view of effectiveness, championed by figures like Les Binet, Peter Field or Byron Sharp. This involves connecting brand-building activities to long-term financial outcomes.

   • Example: A brand manager, instead of merely presenting a Brand Lift Study, now uses a nested MMM to demonstrate how a 2-point increase in "brand consideration" is projected to increase baseline sales by 0.5% over the next 18 months, directly linking brand investment to long-term value.

3. More GenAI applications in marketing measurement. Generative AI is moving from being a novelty to a practical co-pilot for specific, time-consuming measurement tasks, automating data processing and insight generation.

   • Example: An analyst uses an LLM, like in our NotebookLM experiment for the QN#15 edition about Google’s research on MMM, to process and thematically tag thousands of open-ended survey responses about creative assets. This transforms unstructured qualitative feedback into a quantitative variable that can be incorporated into an MMM to score and measure the impact of different messaging strategies.

4. More integrated measurement platforms. The industry is shifting away from siloed tools and toward unified platforms that synthesize multiple data sources (MMM, experiments, attribution) into a single, cohesive framework, as envisioned in the Modern Measurement Playbook.

   • Example: A Head of Measurement no longer presents three separate slide decks. Instead, they use a unified dashboard where the MMM’s top-down ROI is calibrated with recent geo-lift experiment results , and real-time attribution data is used to diagnose short-term performance anomalies within the MMM's strategic context.

5. More "agent-based" automation in measurement. Beyond simple dashboards, we're seeing the rise of autonomous AI "agents" that can monitor, diagnose, and even act on data according to pre-defined rules, reducing the need for constant human intervention.

   • Example: An automated agent is programmed to monitor campaign performance. If it detects a key channel’s incremental ROAS has dropped below a set threshold for 48 hours, it automatically pauses the lowest-performing ad groups and messages the channel manager with a diagnostic summary and a suggested hypothesis to test.

6. Still, most critical marketing decisions will be human-led. Despite vast automation, the final strategic calls—those requiring business context, competitive foresight, and organizational navigation—remain firmly in human hands.

   • Example: An MMM strongly recommends shifting 30% of the budget from YouTube to Paid Search. The CMO, however, aware of a major competitor's upcoming brand launch, overrides the model's output. She opts for a more conservative 15% shift to maintain Share of Voice during a critical period—a strategic nuance the model cannot comprehend and an example of avoiding an "insulting recommendation".

7. Modeling will continue to gain momentum over deterministic measurement. With signal loss from cookie deprecation, URL identifiers, closer scrutiny on local storage, consent requirements or much darker methodologies, relying on observed, user-level "census" data is no longer viable. Modeled and inferred data has become a necessity for a complete picture.

   • Example: An e-commerce brand's weekly report no longer features a metric called "last-click conversions". It now reports on "modeled data-driven conversions," a metric that statistically estimates total conversions by modeling the behavior of unconsented users based on the observed behavior of consented users, providing a more holistic and accurate figure.

8. Marketing Mix Modeling will become a core MarTech feature. MMMs are rapidly moving from expensive, one-off consulting projects to always-on, democratized features embedded within major marketing technology platforms, making them accessible to a much wider range of companies.

   • Example: A mid-sized company, which previously could not afford a full-scale MMM project, now utilizes an integrated MMM tool within their marketing automation suite. Following the path of open-source solutions like Google’s Meridian and Meta’s Robyn, this tool provides them with weekly-updated ROI estimates and budget scenarios.

9. More rigorous testing driven by formal learning agendas. The practice of experimentation is maturing from running ad-hoc tests to executing against a structured, long-term "learning agenda" designed to systematically answer the most critical business questions.

   • Example: A marketing team’s learning agenda for the quarter isn't "test Demand Gen". It's "Determine the optimal Demand Gen spend in combination with video ads in Meta and TikTok." This shifts the focus from tactical wins to durable strategic insights.

10. Growing importance of validated proxy metrics. Given the difficulty of measuring long-term effects directly, sophisticated marketers are increasingly relying on validated, fast-moving proxy metrics as leading indicators of future success. Do not get confused with campaign metrics which have no correlation with effectiveness.

    • Example: An automotive brand's MMM has validated that its "Share of Search" has a 6-month leading correlation with market share. The team now actively uses SoS as a primary quarterly KPI, allowing them to optimize brand-building investment with much greater agility than waiting for semi-annual market share reports.

These ten trends are not isolated events; they all together push measurement in a clear direction. The overarching theme is a definitive shift away from fragmented, channel-specific reporting and toward a holistic, integrated measurement framework designed to guide strategic business decisions.

The challenge for every marketing and analytics leader is no longer just about adopting a new tool or methodology, but about building a resilient and intelligent system and a team that can thrive through this complexity. The goal is not to find a single source of truth (let’s not talk about it anymore!), but to master the art and science of combining these diverse inputs into a coherent strategy that drives sustainable growth.

Industry updates

Google’s Effectiveness Equation

Google has just released a new 69-page white paper on effectiveness. This work has all the ingredients to become a reference today and in the future, the same way that the 2019 paper “Three Grand Challenges”.

This new paper consists of six chapters, covering key areas such as ROI, long-term marketing impact, marketing's effect on price elasticity, smart use of first-party data, effective Marketing and Finance collaboration.

It is a must read for every marketer today so download it and save some quality time to go over it.

Download on this link

Nielsen + Google on the value of AI-powered campaigns

Nielsen has released a case study called The ROI of AI to quantify the bottom line impact of adding AI to campaigns.

It's always valuable when large-scale effectiveness studies are made public, and the recent MMM analysis by Nielsen on Google's AI-powered ad solutions is no exception. The report provides compelling, data-backed benchmarks, suggesting ROAS uplifts of up to 17% for AI-driven campaigns over manual ones. While any robust MMM analysis is a welcome addition to the conversation, it is crucial to consider the context.

The methodology remains unclear. The text mentions “measuring over 50,000 brand campaigns and over 1 million performance campaigns in the U.S. across their AI tools.” This does not look like an MMM or even an MMM meta-analysis. Would love to see more details. The numbers could refer to the whole database but it’s tricky and as a study commissioned by the platform itself, the findings are naturally framed to highlight the benefits of their ecosystem. For marketers, the key is to use these benchmarks not as a universal guarantee, but as a powerful starting point to inform their own rigorous testing and validation.

An alternative to measurement triangle

There have been some public positions by Koen Pauwels and Grace Kite advocating for a new measurement model besides the measurement triangle recommended by Google in 2020.

A full proposal can be read in this November 2024 IPA free report (link). The document is full of sensible, useful and realistic advice, hard to disagree with. However, the news on that proposal is what it’s not included: multitouch attribution. In general terms, while we are fully aware of the limitations of MTA, we would not agree to remove it from a measurement framework altogether. There’s so much detail and nuance in digital media buying that won’t be able to be fully captured either by MMM or experiments, where marketers need insights. Of course one should be aware of its limitations and avoid doing things wrong (like last click attribution), but that is mostly included in existing frameworks of good measurement.

System1 and Effie publish a white paper on creativity

The analysis uses System1 copy-testing methodology and combines it with the huge Effie database. The 41 pages report can be downloaded here and contains multitude of great charts. In particular we loved the following insights:

• Markets that have higher confident on running their discipline use business-oriented metrics, while the ones with less confidence use intermediate metrics.

• The high value of long-running and consistent campaigns

• The importance of “brand codes” beyond the logo (as they say in the report, in other places these are called “iconic assets”. There’s a wealth of assets that can be used such as recognizable shapes, fonts, jingles…

• This paper again keeps challenging the Ehrenberg-Bass school of thought. For all brands, both differentiation and distinctiveness are equally important. However, the value of differentiation (vs. distinctiveness) is particularly high for challenger brands.

Chart of the week

Summer is here! We can’t think of another most relevant topic in this time of the year. In particular, we are having the first heatwave in Spain so we wanted to look at summer-related query trends. This Google Trends chart perfectly illustrates the country's collective mindset as the heat sets in. It's remarkable to see the clockwork-like consistency of these seasonal searches year after year.

As expected, "Playa" (beach) leads the charge, consistently forming the highest peak each summer, closely followed by the general search for "Verano" (summer). Meanwhile, "Piscina" (swimming pool) and "Aire Acondicionado" (air conditioning) follow the same annual pattern, representing the practical solutions Spaniards seek to enjoy or escape the rising temperatures. The 2025 season seems to be shaping up right on schedule!

If we zoom in at queries for “Aire Acondicionado” (air conditioning) and let Google Trends complete the data for June, here’s what you will see:

While the annual pattern is a given, the intensity of the peaks varies significantly from one year to the next, acting as a proxy for the severity of Spain's summer heatwaves.

The data for this last week of June is particularly telling. We are witnessing an almost vertical surge in interest, suggesting that the first major heatwave of 2025 has hit the public consciousness hard and fast.

After months of invitation-only, Google has fully opened Meridian for public use.

Meridian is a framework for building Marketing Mix Models (MMMs). Provided as an open-source Python library, it offers highly customizable options to meet the needs of most companies with various marketing channels. Meridian helps measure the impact of marketing channels, while also accounting for non-marketing variables. Similar to other MMM frameworks, it uses aggregated time-series data on daily/weekly media investment and other media metrics, along with marketing and non-marketing data in the same time-series format, to explain their contribution to sales or other business KPIs.

If you have experience building MMMs and Python knowledge, diving deeper into Meridian is a must for you. The README in the GitHub repository and the “Getting started guide” which allows you to run Meridian end-to-end in a Colab notebook with dummy data, are excellent starting points.

Before starting to get your hands into it or partnering with a data scientist to use Meridian within your organization, we recommend reading further this episode of Quantified Nation to gain background on the research behind it and the ecosystem's response to its launch.

MMM Research at Google: the road to Meridian

Meridian is the culmination of extensive research at Google, focusing on the application of Bayesian statistics to model the impact of advertising on consumer behavior. This involves accounting for prior knowledge, lagged effects, diminishing returns and saturation, and selection bias from ad targeting, which affects search ad evaluation, among other considerations. This is an ongoing research with many interesting additions in the roadmap which will be included in future releases of Meridian. There's a bright future ahead!

In recent years, numerous experts from academia and industry have contributed to the advancement of MMM models in various ways. The rapid evolution of this field is remarkable. Google has contributed to this progress with research that forms the foundation of Meridian. Key research publications include:

• Challenges and Opportunities in Media Mix Modeling (2017 - 3803.pdf)

• Geo-level Bayesian Hierarchical Media Mix Modeling (2017 - 3804.pdf)

• A Hierarchical Bayesian Approach to Improve Media Mix Models Using Category Data (2017 - 3805.pdf)

• Bayesian Methods for Media Mix Modeling with Carryover and Shape Effects (2017 - 3806.pdf)

• Bias Correction For Paid Search In Media Mix Modeling (2018 - 4388.pdf)

• Bayesian Hierarchical Media Mix Model Incorporating Reach and Frequency Data (2023 - 7327.pdf)

• Media Mix Model Calibration With Bayesian Priors (2024 - 7494.pdf)

The four publications published in 2017 (note the consecutive numbering in the Google Research archive) lay the groundwork for the overall methodology. The three most recent publications cover each of the key innovations introduced by Meridian, as highlighted in Quantified Nation #6:

• Control for organic demand by incorporating search query volume data to mitigate paid search bias.

• Enhance video measurement actionability by modeling reach and frequency.

• Improve MMM accuracy by integrating incrementality experiments as prior knowledge.

Recognizing the volume of material, we've created a 27-minutes podcast using NotebookLM, summarizing this research (available also in our YouTube channel). Whether you're new to MMM or an experienced practitioner, we recommend you to listen to this audio. It summarizes the most relevant research Google has contributed to MMM and distributed as a product through Meridian. Enjoy it!

Here’s a quote from the AI generated podcast:

[…] -Speaker 1: The accuracy of your MMM results depends on the quality of the data you feed it. Garbage in, garbage out, as they say

-Speaker 2: Speaking of garbage, one of the biggest takeaways for me is the importance of understanding and mitigating bias […]

Meridian also incorporates insights from other recent research, such as Uber’s Bayesian Time Varying Coefficient Model with Applications to Marketing Mix Modeling (2021) as well as foundational work on causal inference and applied Bayesian inference from Gelman and Rubin’s Iterative Simulation Using Multiple Sequences (1992).

TensorFlow Probability is the responsible for probabilistic reasoning and statistical analysis behind Meridian and it recommends using a minimum of one GPU for optimal performance .

The response of the ecosystem

This project is very close to our hearts, and we recognize that we may have some personal feelings about it. To ensure objectivity, we conducted social listening across various LinkedIn posts, Medium articles, and other sources discussing this Google project and analyzed these discussions using Google AI Studio. While many of the reviews are based solely on the official documentation, it's valuable to summarize the ecosystem's reaction. Here's a summary of the pros and cons:

Pros:

• Bayesian Framework: Enables incorporating prior knowledge, improves stability with limited data, and provides uncertainty quantification.

• Geo-Level Hierarchical Modeling: Captures regional nuances and variations, improves statistical power for regions with sparse data.

• Reach and Frequency Integration: Models ad saturation effects more realistically, potentially leading to more precise measurement of incremental reach.

• Google Search Query Volume (GQV) Incorporation: Provides a control variable to account for external confounding factors impacting organic demand.

• Open Source: Meridian boasts a user-friendly interface that enhances the efficiency and productivity of analytical teams.

Cons:

• Prior Specification Challenges: Subjectivity in prior selection, risk of "same prior, same posterior" if priors are poorly informed.

• Potential for Multicollinearity: Bayesian MMMs can be more susceptible to inflated posteriors under multicollinearity, hindering accurate attribution.

• GQV as a Control Variable: Accuracy depends on GQV reflecting true organic demand. There is also a risk of under-attributing impact of channels that drive search volume.

• Lack of Time-Varying Coefficients: Assumes constant media effects, limiting ability to capture dynamic responses and seasonal fluctuations.

Let’s review some individual examples of comments left, mostly by MMM providers:

• Andy Kozak from Forvio, highlights that current MMMs often treat Reach and Frequency as secondary considerations, and that Meridian will facilitate the measurement of true incremental reach. He also notes key differences between Meridian and Robyn in how incrementality knowledge is incorporated. Furthermore, he points out that if a marketing campaign influences Google queries for the brand—as many do—using query volumes as an estimator of demand risks underestimating the campaign's contribution within the model.

• Aryma Labs published an extensive review in their Substack newsletter, primarily criticizing Meridian's Bayesian approach from an efficiency and almos philosophical standpoint. They argue that prior settings significantly influence the posterior in MMM, which is inherently a "small data" problem. They also contend that Bayesian MMMs are more susceptible to manipulation than Frequentist MMMs, the approach they advocate. This represents a strong contrarian view on Bayesian modeling.

• Kishalaya Mukhopadhyay reports on one of Meridian's key parameters: knots. To capture the effect of different time periods on the measured outcome, Meridian uses "knots" to represent time effects with fewer parameters, rather than assigning a separate parameter to each individual time period. More knots result in a smoother line, effectively capturing seasonality by determining each time period's effect as a weighted average of the nearby knots.

• This eMarketer report on Meridian’s launch provides additional insights into MMM usage within US companies, as well as other measurement solutions. It reinforces the need to combine various measurement methodologies for the most accurate assessment of marketing performance, a much appreciated approach here in Quantified Nation. It appears that multi-channel measurement is a key focus this year, even more than GenAI:

• In case you are looking for a step by step video, Gabriele Franco, the founder of Cassandra (an MMM SaaS solution) has published a tutorial with almost 5K views already, that’s a lot.

• Henry Innis is the founder of another MMM provider called Mutinex. He published an interesting view on the open source vs closed model, where vendors have to protect their IP to innovate and differentiate while at the same time it is needed to enhance trust and transparency.

It is great to see so much enthusiasm about MMMs. We expect it to keep growing as more companies adopt it and there is much more scrutiny on how it is used. There seems to be a great future ahead for Meridian!

Subscribe now

Industry Updates

Felipe Thomaz’s article on multimedia planning

We read Felipe Thomaz' latest article, which was referenced earlier in QN12 and several forums such as this Mi3 Interview. This is supposed to set a new paradigm in the bothism era, after all the Byron-Sharp school's most aggressive claims are facing increasing resistance (for example the renewed interest in differentiation which was anathema until recently). As usual on these academic papers, they are lengthy and rather hard to read, but there are nuggets that are worth following.

This paper is not just “another one”. Because of the gravitas of the authors, the size/complexity of the research and its confrontational nature. This could be part of the corpus of a new era. On the limitation side, the insights are not really "elevator pitch ready" and that will hinder its popularity. In fact, trying to summarize the paper could result in it sounding "old news" where in reality it's not. There's enough argumentation in the first pages on why this piece is relevant vs. the previous methodologies.

The paper analyzes a very sizable database of Kantar's x-media campaigns database which includes 1) Media types and 2) Brand results. What does the paper argue?

First, the paper looks at all the campaigns in the database and compares their results vs. a perfect scenario where ideal results would be obtained. They call it a "stochastic frontier analysis". The results show that 80% of campaigns are able to reach >80% of their "reach frontier", meaning that most campaigns are doing pretty well in terms of reach optimization. However, only 1% of the campaigns are able to achieve >80% of their "brand lift frontier". The author argues this is the fault of mindlessly pursuing reach-optimization as a strategy. That's the starting point and the rest of the paper focuses on 4 brand-lift metrics.

Second, the authors analyze the 1083 campaigns in the database and build 7 clusters of "archetypes", similar to when one does a consumer segmentation study. This is an interesting and useful approach to analyzing campaign types.

Third, there is a long list of results on how the different archetypes and individual media deliver on brand-lift results. Since there are a lot of results, this allows for several ways of interpreting the data.

• As an archetype, the highest results belong to #4 (which is a mix of TV+Outdoor+Facebook+Display). Another interesting one is #3 has both the highest "low confidence interval" and 2nd highest mean result. This means that it's the best proposition in terms of risk/reward.

• Because all archetypes are a mix of lots of media, one can isolate single media beyond the archetypes. In that scenario, YouTube is consistently #2 behind Cinema (which is a really niche media).

One of the authors has been pretty harsh on Byron Sharp publicly several times and this paper would insist in that direction, in particular it criticizes the reach-maximization media planning tactics. In a way, that's the main agenda of the paper. Interestingly, Byron Sharp or his colleagues are not mentioned by name.

Meta Measurement virtual session

Meta will be hosting “Project Robyn & Meta Open Source Measurement Global Community Connect” on February 25th. Registration link

Latest WARC report “The multiplier effect”

A 100 page report with case studies and meta-analysis data that can be downloaded here. Given its ambition, it’s very difficult to summarize. Every marketer and measurement expert will find useful data points, insights and good visuals.

The core message is that Brand and Performance make sense in an integrated way: there should be no “brand campaigns” or “performance teams”, because they both work together and in fact reinforce each other. A strong brand will make the performance marketing results even better and also help to command a price premium. This goes very much in line with the growing consensus on how advertising works, beautifully summarized by Les Binet. What readers of the report will find is even more data points on why it’s the case.

In what regards to measurement, the paper takes a hierarchical approach on the measurement triangle (MMM, experiments, attribution) and puts MMM in the most important position, with experiments and attribution as complementary.

Chart of the week

Spring is coming soon! It’s remarkable that there is some consistency across years and across countries with such a different climate!

Oldies but goldies

Today we want to bring an old book which ideas remain in marketers minds for a strong reason. "Scientific Advertising”, written in 1923 by Claude C. Hopkins. Available for download here

This book advocated for a shift from creative expression to data-driven salesmanship (we know both are needed), pushing the idea that advertising should be treated as a quantifiable investment, demanding measurable results (all in!). He started techniques like couponing and testing in direct mail to track campaign effectiveness, establishing the basis for modern direct marketing and testing.

Furthermore, Hopkins stressed the importance of understanding consumer psychology, urging marketers to present their messages from the customer's viewpoint. He advocated for specific, factual claims over vague assertions, believing consumers respond better to concrete information. The power of compelling headlines, designed to draw readers into the full advertisement, was also a key principle. Look at who wrote the introduction to the 1960 edition:

Nobody, at any level, should be allowed to have anything to do with advertising until he has read this book seven times. It changed the course of my life.

David Ogilvy

This book is a great read for every marketer who wants to make data driven decisions and enjoys understanding the history of it, promoting an analytical approach, measurable results and deep consumer understanding. No doubt these are principles that are still highly relevant in today's marketing landscape.

Thank you for reaching the bottom of the newsletter, share and see you soon!

Share Quantified Nation

Furthermore, Economic theory offers valuable insights into how we make decisions. Marginalism is a great example of these. Ultimately, you want your marketing budget to be spent where it will generate the most revenue, correct? Therefore, you should aim to spend on every channel and strategy up to the point of equal marginal ROI. Before you rush to enroll in a microeconomics course, read on!

Subscribe now

How to get abnormally good results?

The simplest economic models of a company tells us that in the short term, a given company will produce as many products as long as the marginal cost is lower than the marginal revenue, to the point where they equate each other.

That situation would generate “economic profit” (details here) which is a desired but unattainable situation in the long run, in perfectly competitive markets.

Prompted by those amazing profits, more companies enter the market, the price goes down and our initial company will produce a quantity to sell at a price that enables to earn a “normal profit”.

This simple model works under the following circumstances:

1. Homogenous product

2. Great quantity of buyers and sellers

3. Absence of barriers of entry and exit

4. Companies aim to maximize benefits

How to get abnormally good results? It’s almost impossible

One way to approach this, is to look at deviations from the preconditions of the model that leads to normal profits. Let’s take 2, 3 and 4 for granted, they talk about external factors often out of our control.

Instead, lets’s focus our attention to 1, with the broad lens of the 4Ps: product, price, promotion, place. If our company is able to justify a meaningful difference in one of those 4Ps, we are in a better shape to sustainably get abnormally strong results.

However, how likely is it? For the very vast majority of companies, not that much.

• Product: Most often the products are fundamentally equal to each other: a shampoo or a detergent will get the job done, in a similar way, for the vast majority of consumers. Of course the brand and the performance of the product would be different in the degree of how well they perform the job, but not in a fundamental way. That’s why products compete with each other, because people consider them valuable alternatives.

• Price: again there are differences, but most often the competitors operate under the same principles. Selling at 3€ vs. 2€ does not really make you fundamentally different. What would be different? The Nespresso model of machine+capsules was different for many years, the Spotify music subscription model was also different… Having just a slightly different price point, is not really a difference.

• Promotion: most companies have similar marketing channels and media strategies to each other. Of course some do better than others, but that’s not sufficient to generate relevant “economic profit”

• Place: the healthiest strategy here is to be in all places where shoppers are, which leads to no differentiation. Can you be creative? Of course, like all those D2C companies that tried but ultimately resorted to selling wherever shoppers are.

So, for the vast majority of companies, “economic profit” is a quimera in the long term, because we operate in competitive markets and we are not fundamentally different from our competitors.

So, what’s left? Doing a little better, everyday

It’s all about doing the best we can with the resources we have, trying to improve a little bit everyday. Will your a better MMM process and methodology give you a competitive advantage? It may do in the short term, but not in the long term. Nevertheless, you have to do it because that’s what best-in-class companies are doing, and it’s just needed to stay in the race.

That’s the game we are in. The game of endless optimization. The race can’t be “won”, but that’s not the point. We just need to keep moving.

Industry updates

The hidden 50% of your ROI

Google has published an article with solid datapoints on long term ROI multipliers and brand vs. performance budget allocations. These are based on MMM projects across different research vendors, it's a great starting point for strategic 2025 marketing budget discussions.

The Python version of Project Robyn is open for beta testers

Up until today, using Robyn was almost equal to using R. This has changed! As Meta has announced a a Python version of Robyn. This was possible thanks to AI, as they state in the page:

The R-to-Python porting was completed leveraging the latest Large Language Models (like Llama) and open source AI frameworks, demonstrating both the power of AI and our ongoing commitment to equip entrepreneurs and advertisers with cutting-edge, best-in-class measurement tools.

A new library for calculating LTV

The Meta Marketing Science team has released a new library for calculating LTV which will be useful to feed bidding algorithms with metrics beyond revenue. These initiatives are key to ensure optimization towards the long term profitability of a business.

The problem on LTV vs. effectiveness is rooted in the inability that we have to accurately and reliably measure the long-term impact of marketing campaigns on revenue, not to say in customer lifetime value. Smart marketing and branding efforts play a crucial role in boosting LTV by building strong customer relationships, increasing brand loyalty, and ultimately driving repeating customers and higher value over time.

Aren’t you a Mark Ritson follower?

Some Mark Ritson best moments in YouTube, when he discussed about the moment he commented a post on LinkedIn telling the author that the model “sucks ass, basically, right? please go and read more and stop doing this”. This interview for the Marketing Architects podcast is a great introduction to Mark Ritson in case you do not know him.

And of course the Mark Ritson's Top 10 Marketing Moments of 2024:

Two great videos to upgrade your marketing skills!

Chart of the week

Con mi burrito sabanero voy camino de Belén 🎵 🎶… Tuqui, tuqui, tuqui… tuqui, tuqui, ta… 🎵 🎶 Depending on where in the world you live, this has been part of your Christmas soundtrack. Google Trends certifies it, highest ever interest!

As a musical closure, here the latest version by David Bisbal, it is never too late to close Christmas. (Did we say happy new year?)

• El proceso de creación de logo, colores, animación, introducción del podcast que hicimos con Toño Domínguez

• El nuevo cuadrante de Gartner sobre proveedores de MMM, no es oro todo lo que reluce

• La nueva presentación de Benedict Evans y qué contiene de medición de marketing

• Nuestro apoyo a Dadadata con su campaña para conseguir fondos para Valencia

• El último chart of the week del año, con unas visualizaciones de Carlos Real

¡Hasta el próximo año con más Quantified Nation!

When building brand assets, you need both consistency and adaptability to different contexts. A brand should aim to outgrow its creators, so it's essential to be ambitious and forward-thinking. Trust your gut, too. Early-stage projects often lack the robust metrics needed for data-driven decisions, so you'll need to rely on past experience, general knowledge, and a dash of data to guide you. You definitively need someone to help you with that brand creation.

We tried searching for professionals on online platforms without success—or perhaps we didn't try hard enough. Luckily, Toño Domínguez (LinkedIn) agreed to help us on this project. He guided us through a fast-paced yet thorough process, which we wanted to share here in this post. If you like his work, let him know and share this with others! He has a great methodology.

We're also currently looking for someone to handle video production. If you know anyone, please let us know!

Subscribe now

The process

Toño has excellent drawing skills. We initiated the process by sketching different ideas...

Among all these ideas, there was one which we found intriguing. We liked the concept of a measuring tape, which is obvious as the topic for this newsleter is mainly Marketing Measurement. But also reminds of a Q and it can expand and contract…

So the main idea for the logo was settled. From here, Toño was able to develop and connect additional concepts to suggest the logo from where he would expand to the additional assets.

The main idea of the measuring tape helped him find not only a suitable color palette but also the whole challenge of converting a measuring tape in a brand. Which he did great!

And started playing out with options…

The final outcome

We ended up choosing which one we considered more polished and aligned with the different environments where this brand should be seen.

What else?

Having the logo, colors and style, he created an animation which would work as an intro and outro for the podcast:

The second part of the animation includes our images and the episode number, which we can edit ourselves using Canva.

This animation also needed some music. Well, Jesús recorded himself with a digital accordion playing 20 seconds of Surf’s Up, by The Beach Boys. Here’s the result:

Check the final result in the last episode of the podcast or in our new YouTube channel: Quantified Nation

Subscribe to our YouTube channel if you want to hear us in Spanish. There’s always subtitles so you have no excuses.

Industry updates and upcoming events

1st ever “magic quadrant” for MMM vendors

For the first time, Gartner has published a quadrant assessing MMM vendors. As one could expect, the “winners” are widely publicising it in their own media channels. Our view is that the assessment criteria are very much infrastructure related, it’s like Gartner is judging a software plaform. It’s great that MMM gets into the spotlight, but we doubt this particular work is a useful assessment. You can find full details on this link

AI eats the world

Not purely marketing measurement related, however everyone working in tech related roles should be aware of this presentation. Coming from a long lasting tradition of summarizing the greatest macro and strategic trends in technology, this year’s Benedict Evans presentation focuses on AI. It provides many insights backed with data so it is worth mentioning.

We have discussed about the relevance of AI for marketing measurement and some practical applications like pretesting or data exploration. It is an intriguing area for many and we are all trying to make it useful for different purposes. As Ben says, we are in the context of “Huge interest, limited use so far”.

It also includes some insightful data about the advertising industry, where Amazon is starting to play a bigger role by letting brands pay them to influence their decisions through advertising. It seems it is working!

Also, media revenue is quite widespread across companies. While the ecosystem is different depending on the region or country, measuring lots of advertising providers makes it harder to have a full picture of how to reach this population with ads.

Are you worried about showing your ads too many times?

Frequency management is a challenge for many brands. The Marketing Architects Podcast has recently launched a new episode on ad fatigue and frequency management.

They share interesting conclusions such as “The first time someone sees your ad drives the strongest response.” Further impressions always have diminishing returns. Video drives less fatigue but frequency management is complex due to media fragmentation (see above). Many additional insights are discussed so check it out!

Supporting Dadadata and Valencia crisis

The newsletter Dadadata is donating all subscription revenue to help those affected by the crisis in Valencia. This is a great resource for Spanish-speaking professionals in the digital tracking field. Subscribing means you'll develop your skills in a crucial area and, most importantly, directly support a community in dire need.

We would like to collaborate with them writing an article. What kind of content would you like to see to encourage your subscription? Let us know in the comments

Chart of the week

2024 is over and Google is issuing their yearly trends report. Our friend Carlos shared the following visualization of the Spanish top trends using this R package

We hope you have a very well deserved holidays and enjoy some time off. We expect 2025 to be very busy for Marketing Measurement so get prepared for more Quantified Nation.

• Cómo hacer que las organizaciones los utilicen para la toma de decisiones

• Factores clave para el éxito: involucración de las personas adecuadas, convertir proyectos en procesos, asegurar la participación activa

• Entender las limitaciones de los MMMs y acercarlos a la realidad

• Novedades sobre el debate entre Felipe Thomaz y el instituto Ehrenberg-Bass sobre la disponibilidad mental

• El nuevo informe de WARC sobre la relación entre métricas de marca y de ventas a corto y largo plazo

• NielsenIQ lanza servicios de MMM

Además, estrenamos portada y nuevo branding.

¡Esperamos vuestros comentarios!

• KPIs mal alineados, incentivos perversos y enfocarse en lo que puede no es importante para el negocio

• Adquisición de Magic Numbers por Analytic Partners

• Partnership de Effie con System1

• El debate abierto sobre si la atención influye en la eficacia publicitaria

¡Esperamos que os guste!

For those who have run these projects before, the harsh reality is clear: driving change is hard. Expensive and exhausting MMM projects often lead to extensive debrief sessions but ultimately lose momentum. Months later, how much real change do these projects deliver? Why do they fail to "move the needle"?

Subscribe now

The soft element

Effective researchers need to recognize that they are in the business of driving change. It’s not just about the results—countless articles and books have already covered that. In this newsletter, we aim to highlight some elements that have worked well in our previous projects:

• Having the right stakeholders and sponsors: This means not just sufficient seniority but also ensuring a broad representation across teams.

• Asking the right business questions: Collect business questions but avoid a "shopping list" approach. You may find that your stakeholders think in different terms, but MMM is often flexible enough to adapt its modeling to answer the key questions.

• Shifting from a “project” to a “process” mindset: A project gets delivered, and people move on. A process is ongoing, with research serving as one component of a continuous loop.

• Making stakeholders “sweat for it”: Encourage stakeholders to contribute data and business context. This not only leads to better outcomes but also ensures they have skin in the game.

• Avoiding “insulting” recommendations: An “insulting recommendation” is one that, even if supported by data, is too detached from reality. It can make you appear like a brilliant mathematician disconnected from the business world.

• Precision is not the ultimate KPI: The goal of MMM is to drive ROI improvements, not just to create perfectly precise explanations of past performance.

• Allowing enough time for discussion during debriefs: Obvious but often overlooked, robust debate is critical for gaining alignment and ensuring actionable outcomes.

The hard element

Good recommendations must be translatable into actionable steps for the people implementing them. Often, MMM outputs are difficult to translate into clear, practical decisions. What do we mean?

• Budget shifts are not simple: Today, a large portion of media purchasing is automated. For instance, increasing or decreasing investment in generic paid search requires navigating multiple options. Additionally, MMM models rely on past performance, but automated media buying adjusts dynamically. Changes in how competitors bid for the same keywords can drastically alter response curves.

• Small changes often matter most: Drastic changes based on MMM outputs can invalidate the model itself, as the context in which it was built shifts. Test large changes in a controlled environment to validate the direction suggested by the MMM.

• Look at further implications of the changes: A common MMM recommendation might be to reduce GRPs per week and redistribute them to avoid blank weeks. However, this can reduce Share of Voice (SOV) to levels that media agencies and brand managers might consider unsafe. Understanding these nuances is critical.

These are just a few examples on why making the most of a MMM is far from guaranteed and is often not achieved in the first iteration. Resilience and iteration is a must when a project seems to be failing. In marketing measurement as in many other fields, failures must lead to learnings and future success, or total ROI improvements.

MMM innovations may be of help

Methodological improvements are intellectually satisfying but often result in only marginal increases in a project’s real-world impact. However, some improvements are particularly valuable—those that close the gap between the model’s outputs and actionable steps.

For example, modeling Reach and Frequency and going beyond Impressions. Media planners and buyers often work with reach and frequency metrics, so presenting results in impressions (even if converted to GRPs) creates an unnecessary barrier to action.

Traditionally, MMM outputs have been hard to navigate, static, making it challenging to update findings with new data. Today, many providers offer interactive dashboards that integrate data from multiple sources, update results regularly, and provide forecasts and scenario planning. For many organizations, this shift makes MMM outcomes significantly more actionable.

The success of MMM projects doesn’t rest solely on the quality of the model or the accuracy of the data. It relies on the ability to drive organizational change and translate recommendations into actions that stakeholders can realistically implement.

Industry updates and upcoming events

Is pursuing mental availability a safe bet?

Felipe Thomaz has openly critized Byron Sharp’s insights in a recent interview in Mi3 (link). He has conducted a recent research with Kantar and Wavemaker that, according to him, proves that pursuing mental availability through mass reach is not a safe bet. The research full research will be published in 2025

New WARC report on “The holistic path to measuring media investments”

Google’s Michal Protasiuk and Ahmet Bas have published a recent report in partnership with WARC, based mostly on Nielsen MMM (with some Ipsos MMM, too). The research quantifies the link of brand equity metrics with short and long term sales. It also reconfirms the 60/40 rule of branding/performance. This WARC report is open to the public and you can read it on this link.

The next chapter of the measurement providers saga

Interesting move on the MMM-vendors space: NielsenIQ has launched MMM services. The evolution of the Nielsen saga is pretty complex. What was one company, in 2020 got split into Nielsen Media and NielsenIQ. At first, Nielsen Media kept the MMM capabilities and NielsenIQ focused on sales/household panels. Later NielsenIQ purchased GFK, also rebranded to NIQ. In the meantime Nielsen Media rebranded itself as “Nielsen”. In this latest move, NIQ enters the MMM space, effectively becoming a competitor to Nielsen Media. More details on this press release.

Do not miss: Meta EMEA Measurement Forum

Meta has recently held their 2024 Measurement Forum in EMEA. This is a strong signal of commitment of Meta towards measurement in general and MMM in particular. You can check the recordings here. It’s clear that Long Term ROI and brand building are coming into the mainstream. They also share a Kantar meta analysis on creative best practices (in QN10 we discussed about that topic).

Chart of the week

Are you into Padel or Pickleball? Both of those sports are exploding in popularity, with a really interesting geographical divide.

Oldies but goodies

Les Binet is the Group Head of Effectiveness at adam&eveDDB and has become an essential figure in marketing effectiveness, challenging marketers to think differently about the balance between long-term brand building and short-term sales activation. He is also very active on LinkedIn where through his profile he shares very insightful content about effectiveness. His work with Peter Field, captured in The Long and the Short of It, exposes a critical tension: the constant need for immediate results versus the patience required to build sustainable growth.

The essence of Binet’s message is clear: short-term activation campaigns (promotions or performance marketing) can deliver a quick win, but they rarely strengthen a brand in the way long-term efforts do. Brand-building campaigns, designed to evoke emotion and create lasting associations, take longer to pay off but are essential for profitability in the long run.

This is a conversation we have held with many marketers, often agreeing the following: because short term results seem easier to measure and can be reported to management in an easy way, companies have a strong bias towards them. This generates an increasingly higher and long lasting dependency on performance marketing. Also, these companies have fallen into a trap which is very hard to escape from.

The effectiveness data in the book are derived from the IPA Effectiveness Databank – the product of 30 years of the IPA Effectiveness Awards covering more than 700 brands in over 80 categories.

For those of us obsessed with measurement, this creates a challenge. Tools like attribution models and performance focused MMMs are often better suited to measure short-term performance, reinforcing a bias toward immediate ROI. But over-focusing on the short-term risks trapping businesses in a spending treadmill, where results disappear as soon as the budget is paused. At the same time, ignoring short-term performance can lead to impatience and missed opportunities.

Binet’s principles are more relevant than ever. The question is: how can we measure and balance both horizons? While data-rich environments make short-term results easier to quantify, we must push measurement tools to accommodate the bigger picture, too. That’s where frameworks and discipline on the measurement methodologies can help. Not just to measure what’s working now, but to guide sustainable, long-term growth. Data from the research also shows that in order to obtain maximum effectiveness, “brands should spend around 60% of their budget on brand-building activity and 40% on activation”

In previous editions of Quantified Nation we have discussed about branding metrics and how to link them into business results or the value of nested MMMs, which capture better the long term effect of marketing. Also, Share of Search is suggested as a proxy for share of voice to understand the degree to which a brand is top-of-mind for consumers.

There are many learnings in this book with valuable examples, research results and easy to digest conclusions. Go ahead, send it to your e-reader, get a copy or print the pdf. You will always be in a better position as a marketer if you read this book.

• La creatividad como elemento fundamental del marketing

• Cómo incorporar elementos de la creatividad en la medición

• Los updates sobre third party cookies de julio de 2024

• La publicación del playbook “Modern Brand Measurement”

• El chart of the week de los Juegos Olímpicos

• Un comentario al ya clásico “How Brands Grow” de Byron Sharp

• El enfoque frecuentista vs. bayesiano en marketing.

• Herramientas de código abierto como PYMC.

• Eventos como la conferencia de ASI sobre medición televisiva.

• Pre-testing de campañas con IA.

• Investigación B2B con consumidores virtuales.

• Comentario al artículo "Probability, Frequency and Reasonable Expectation" de Artie Cox.

Esperamos que os guste!

It is very common to associate the abundance and availability of data and metrics with data-driven marketing. This is not true in most cases and the paralysis by analysis example is one of the most popular illustrations. Savvy marketers learn how to make decisions with insights derived from the data. They understand which metrics to look at within the sea of available data when there is a performance issue or an unexpected outcome. But they also tend to focus on few important metrics and look at them with a business lens.

In this context of data abundance, we need people who have a bias towards action. Most of the time, this is for the better. More actions provide more data, giving those hesitant to make decisions more information to eventually make the right choice. However, it is easy to fall into different traps where being data-driven does not lead to better decisions. We all have experienced them, but in most cases don’t know what to do to avoid them, mostly when leadership wants to show they are data-driven. Here it goes:

1.- Focus on what can be measured instead of what is important

Being data-driven often leads to focusing on metrics that are easy to measure rather than those that are truly meaningful. By contrast, those meaningful are harder to obtain (incrementality metrics are a great example of this one).

It is heart breaking to see how marketers use today metrics like impressions or clicks as the goal of their campaigns simply because they are readily available, while ignoring more impactful indicators that require deeper analysis (reach, viewability, qualified visits,…). This focus on what can be measured, rather than what is important, leads to suboptimal decisions by missing crucial information.

We all can think of examples where certain initiatives or strategies are not considered not because there is no expected business impact but because they do not impact on the metric you are asked to optimize. Imaging having to optimize paid social media towards a last click because the source of truth uses a cross channel attribution solution which gives credit based on last click. You are optimizing social media ads in a way that is missing the whole potential it can have in upper parts of the funnel just because it is harder to measure.

2. Illusion of control

When surrounded with data everywhere, you might have the impression that everything can be measured and managed. In reality, marketing is influenced by countless unpredictable variables. Over-relying on simplistic data can create the illusion that you have control over outcomes, which can lead to frustration when the numbers don’t align with expectations.

Instead, assume that reality is complex and through measurement we want to understand it a little bit better each day.

True control in marketing also requires intuition, creativity, and the ability to adapt when faced with unexpected challenges. A balanced approach which includes both data and human judgment with a growing understanding of how marketing works can help navigate these complexities more effectively.

This illusion of control is also abundant when a senior marketer is asked about goals, targets and measurement, when defining plan and the numerous KPIs that will be used to monitor performance. It is not about what you know you can measure

3.- Take credit when metrics go well. Blame external factors when metrics go bad

In most of the cases, success is attributed by implying causal relationships. Authority bias plays a fundamental role in establishing these relationships. However, when metrics decline, the focus shifts to justifying why external factors are responsible such as changes in the market, economic conditions, or product updates, rather than examining potential internal improvements.

This context makes it harder for the team to draw unbiased connections between the data and the actions taken. By avoiding accountability when things go wrong, teams miss out on opportunities to learn and improve. True growth comes from analyzing both successes and failures, understanding what truly drove the results, and making informed decisions based on those insights.

4.- The trap of Constant Optimization

With data available as a stream of water, marketers are often driven to continuously optimize every detail of a campaign. While optimization may seem beneficial, an overemphasis on small, incremental changes can shift attention away from strategic initiatives that truly drive growth.

This constant focus on tweaking minor elements often results in diminishing returns and poorer decisions, as the team loses sight of the bigger picture. Instead of investing time in bold, impactful strategies that could differentiate the brand, marketers can be busy with adjustments that fail to deliver meaningful business impact. Ultimately, this approach reduces the overall effectiveness of marketing efforts by optimizing towards what is easy to measure, usually short term goals.

5.- Data-driven can lead to less creative outcomes

Data can offer valuable insights, but it can also hurts creativity. This is because the data supports repeating successes of the past, rather than trying new ideas. This focus on the numbers can stop people from taking risks, making it hard to create unique marketing strategies that stand out.

Data is a powerful tool and we love it. However, like any tool, it's important to understand its limits and potential downsides. Being aware of these challenges can help you find a more balanced approach, one that allows for creativity, connection, and sustainable marketing strategies.

Marketoonist, as profound and true as always gave us these 7 deadly sins of data-driven marketing:

What to do now?

For marketers, the key is to use data as a support rather than as the sole driver of decision-making. Focus on the metrics that matter, but don't lose sight of the fundamentals of marketing, specially brand building and long term effects. Balance data-driven optimization with strategic, big-picture thinking.

This also means ensuring that marketing teams dedicate time not only to analyze and review insights from data but also to creative processes, moments where ideas are shared freely without the immediate constraints of metrics. For instance, marketing teams should set aside time to define qualitative long term goals and assess as a group if the marketing activities for the period are aligned with these long term goals.

In addition to these team dynamics, when working with agencies or partners, it’s essential to clearly communicate both the data-driven goals and the brand's larger vision. Data should be used to provide direction, but agencies should also be empowered to bring new ideas to the table. Encouraging partners to take calculated risks without fearing failure can lead to more innovative campaigns that truly resonate with audiences. However, giving an agency the ultimate goal of optimizing ROAS measured by a platform is a poor definition of performance and sustainable growth

Having solid marketing foundations and knowledge about how consumers behave are essential to ensure these risks are calculated and directed towards improving business value. Blind test and learn or a pure attitude of “let’s test everything” only leads to unnecessary costs and inevitable failure.

This balanced approach will lead not only to better results from your marketing and sales efforts but also to more meaningful, long-term growth. Think of great brands such as Coca-Cola and Apple. They use data to understand customer behavior but don't shy away from investing in creative branding campaigns that build emotional connections with their audience. They have learned that while data can guide marketing, creativity and strategic risk-taking are what truly differentiate a brand.

Industry updates and upcoming events

Analytic Partners acquires Magic Numbers

Analytics Partners is one of the largest providers of marketing measurement, including MMM. Magic Numbers is a consulting company composed of “Thoughtful business economists who are good with numbers and understand how consumers and markets behave”. It’s amazing to see some M&A activity in the media effectiveness measurement sector. We could be entering a new moment in the market where this happens more often.

Announcement from System1 and Effie Worldwide

System1 has announced (link) a partnership with Effie Worldwide to research the field of creative effectiveness. System1 already has an extensive track record with UK’s IPA and is probably planning to do a similar approach with the 165,000+ global ads from the Effie database. The timeline? The next two years is to share practical advertising insights for sustainable brand growth. Especially for marketers in the US.

MMM Best Practices and Success Stories with ARF

The ARF hosted a session with Google’s MMM experts (link) on September 10th. A summary can be accessed by ARF Members on this link. The core message was on achieving higher levels of actionability in MMMs thanks to more granular data, incorporation of metrics such as reach/frequency, priors… and cooperation between clients and modeling agencies.

Learnings from the analysis that AI did to over 8,000 top ads

As we discussed in QN10, a good creative is the biggest driver of effectiveness. In this recent Google research (link) you can see some insights and best practices. It is to be noted that the research was conducted to explain views/likes, not final business outcomes.

The drivers of out-of-home creatives

System1 has analyzed the drivers of good out-of-home creatives and published results on this column and this downloadable whitepaper.

Latest marketing debate

Do Attention Levels Drive Ad Effectiveness? Byron Sharp Says No

There is an increasing interest in the concept of attention in the marketing and effectiveness world. Proponents say the level of attention and the attention span is critical to have an impact. In this Forbes article, Byron Sharp claims for the opposite, saying that an excessive focus on attention metrics can be misleading for advertisers.

Sharp argues that while attention is important, “more attention does not simply translate into greater effect”. For instance, optimizing solely for longer attention may lead to higher costs and limit overall reach, while ignoring the importance of effective frequency and broad exposure. “Yes, ‘Mere exposures’ can do good, even if they are fleeting exposures, which is the case for most advertising”, enhancing the mental availability of the brand being advertised.

Sharp recommends that advertisers focus on creating memorable campaigns that leverage distinctive brand assets to achieve broad, consistent exposure, even if individual impressions are brief. This perspective aligns well with the idea of balancing data with creative strategy to ensure that marketing efforts drive real, sustained growth.

This can be another example of focusing on what can be measured instead of what is important so it is a good closure for this edition of the newsletter.

Chart of the week

Electric bikes are really rising popularity at the expense of traditional mountain bikes!

Let us know what you think by replying to this email or commenting in Substack. Share this newsletter with your competitors so you can level up your industry.

• Data-driven marketing, cortoplacismo y mediciones complejas

• El projecto Halo de la WFA (World Federation of Advertisers) para medición cross-media

• Nuevo estudio de Kantar “People-Powered Measurement”

• Oldies but goodies: Decoding decisions cumple 4 años