QuantifiedNation #3: KPIs and attribution, what else?
Food for thought
Key Performance Indicators (KPIs) are the important metrics you care about in your business. When you bring the concept to marketing measurement, those should be the main indicators of the marketing performance. A measure of goal attainment. The number showing how far or close your marketing activities are from the objective. A KPI is not a metric (visit) but a metric with a context (visits per channel since the beginning of the quarter compared with the same period the last year) and a target (+5% on average by the end of this quarter). That is, a metric might be “number of website visits”, but a KPI adds context and target, like “increase visits per channel by 5% compared to last year's quarter”.
Once these are established, they serve as the north star to which everyone in the organization must be aligned. Top down and bottom up. The right KPIs are like a well calibrated GPS so you not only know you are moving but you know you go in the right direction. Good KPIs ensures the team is busy doing the right things and foster a sense of ownership and responsibility. Helps you allocate resources and provide a common business language which improve the communication of goals, progress and achievements.
KPIs require frequent target setting, so everyone can see how far is the star. Target achievement must be a relevant part of the salary for everyone working in marketing with consequences when not achieving targets consistently. Overachieving targets on KPIs are also a great way to justify good performance in annual reviews and get that promotion with objective statements.
In many cases, a poor understanding of the different data sources as well as not caring too much about the measurement strategy leads to a team with the wrong KPIs, with terrible but often subtle consequences. However, setting the right KPIs and targets is an ongoing process so regardless where you are, there is always room for improvement. KPIs also evolve over time to reflect changing business goals or market conditions so make sure you have good habits for KPIs setting.
It is difficult to have a common framework for marketing KPIs. Nevertheless, the association of Swedish Advertisers (Sveriges Annonsörer) has launched in October 2023 the Effectiveness System 2.0, proposing a framework with 36 KPIs to measure the impact of marketing investment. The first version won the President’s Award by the World Federation of Advertisers so its worth taking a look at it, here’s the summary:
Digital metrics, KPIs and conversions
Let’s start with digital metrics. Digital businesses have a strong attachment to website (or app) KPIs, specially around the concept of conversion. Digital conversions are key behaviors in a website or app, such a purchase, a lead generated, a subscription or any other interaction that the marketing team might consider a as a goal. These are collected by tools such as Google Analytics by tracking user behavior and attaching such behavior to an ID. This ID is most often defined through a first party cookie in the browser of the user, provided that the consent has been given to such data collection. No consent, no cookie, no ID, no conversion.
This data source is essential for understanding the performance of your marketing activities that happen within the website or app such as promotions, creatives or an improved customer experience. However, for those marketing activities happening outside the website or app, such as digital advertising or content creation, this data source becomes flawed, as the full journey of the user cannot be associated with confidence to an ID.
Let’s look at the following example: if you use a metric such as “new users” for one of your KPIs, it will be normally defined by the number of visits to your website that did not have an ID (a cookie), because it was the first time they visited with such browser, cookies were deleted or because it was the first time the user consented to data collection. You will need to consider to what extent such metric is a good one for your KPIs and how you should handle it. You could be having great numbers on new users from a certain campaign just because it is being opened from new devices, not from new users.
Attribution and third party cookies
In the past, third party cookies allowed advertisers to attach an ID to every user and have this user identified wherever an ad could be shown and most consumers only had one device. Companies collected data on every marketing impact of every user through technologies that allowed them to identify the user until the final conversion. This use case for cross site tracking led to the common practice of building attribution models considering the full digital user journey, from the ad impression to the conversion. Today, third party cookies are not available anymore, being Chrome the last one to remove them in 2024. For this and other reasons, full funnel attribution is an unrealized promise of digital marketing measurement. Forward looking solutions do not seem to have the chance to recreate such customer journey. A frustration for many, an opportunity for those who embrace other methodologies or data sources for their magic KPIs.
Yet, many companies still use KPIs with metrics such as new users, attributed conversions, unique visitors and others from these sources beyond its possibilities. This leads to setting the wrong incentives, a difficult scenario for cooperation or the ability to trick the metric with ease. Surprisingly frequent.
Let’s take the example of a company which collects leads through the website and convert those leads into customers through a sales process. This company will generate leads through Organic and Paid Search, Affiliates, Social Media, Display advertising and many other channels. They will know what is the traffic source for each lead generated. Then, they trace leads until they become customers and report sales per marketing channel.
You compare these sales with costs per channel and you will get Cost Per Acquisition. Alternatively, if you use the value of each sale, you will have ROAS (Return on Ad Spent) per channel. CPA and ROAS will allow you to make smarter decisions on where to put your budget in the next period. If your company does not collect leads online but sells online, it is very likely that you get metrics like ROAS directly in a dashboard, using similar data sources.
You will end up having conversions per channel, CPA and/or ROAS in a report to make decisions. However, you will be looking at numbers per channel based on which brought the user to the website when converted. That is, a last interaction attribution model or similar.
In many cases, such reports are sent to the finance teams so they evaluate the performance of the digital marketing strategies, trying to understand why marketing wants more budget. Furthermore, you can work with an agency, an affiliate or an ad technology that gets paid based on such attribution.
Just because it is available and it has been a common practice in digital, it does not mean it is the best option for your decision making. You can do better.
The case for conversions and attribution in KPIs
Website or app data, mainly in the form of conversions, is a great asset for optimizing digital advertising campaigns where users frequently click on, such as paid search and to some extent, paid social. Optimizing towards attributed conversions or attributed conversion value is a vast advantage given current development of solutions using AI to optimize ad performance. This is even more powerful when we allow the solution to look at all “known” journey using a data driven attribution model, not just at the last click.
Even when using such type of attribution models, conversions only serve as a guide to understand a small portion of the customer journey. It can be a problem if you believe this is the portion where you should only focus on. You will end up capturing existing demand with unsustainable costs.
Using these KPIs will bias you towards making decisions based on what happened after a click, losing most of the impact that the majority of marketing channels bring to your business. Also, you are making decisions based on how channels compete for being the last click before the conversions, setting the wrong incentives for the teams or partners handling these channels. In order to get the most of your marketing investment, you need to set the incentives for your channels towards cooperation, not competition.
Be mindful of the KPIs you set for your marketing activities, give credit where the credit is due and avoid data sources that sabotage your growth. Ensure the KPIs of your marketing channels do not rely only on digital conversions, much less if it uses a last interaction model or do not have modeled conversions. Triangulate with experimentation and mix modeling to ensure you make the best marketing decisions.
Quantified Nation brings you Marketing Measurement Magic through the fingertips of Jesús Martín and Pablo Pérez. Subscribe so you do not miss future editions and help us spread the word for a better Marketing Measurement world
Hot Takes
Budgeting solutions
Looking for inspiration on building your media plans for 2024? Here you have two open tools based on benchmarks (UK) to start the conversation. Very different approaches on each, it’s great to see how MMM-based models can have vastly different outputs.
Magic Numbers “Recovery Budget Planner” focuses more on projections of revenue and profit, the kind of metrics to drive a good CEO-CMO-CFO conversation.
Thinkbox “Media Mix Navigator” for more details on the media mix
Increase your knowledge on Marketing Mix Models
Long read published in ThinkWithGoogle in partnership with Hakuhodo DY Group. >130 pages of really especific content on making better MMM. A truly rare read, lot of care on lenght, details and discussion on practical challenges. Go and dowload it!
Measurement and business growth in 2024
Google article “3 key measurement considerations to drive business growth in 2024” (link). Short read with thought starters for your 2024 measurement strategy.
Become a rockstar advertiser
Great read from Nigel Hollis on “Becoming a rockstar advertiser” (link) that includes data-based recomendations on brand building, media mix and creativity best practices. Comprehensive compilation due to its breadth and the quality of sources cited.
Mix Modeling is more about people than about models
Automation capabilities and open-source models are one of the reasons behind the rise of MMM popularity. However, all models require a great mix of human+machine, art+science, experience+data. 17 measurement effectiveness experts have signed a letter with some warnings (link to letter).
Embracing new techniques is usually a good thing and automated models will be continuously improved to solve for their most obvious challenges, also becoming more flexible and less “one size fits all”. However, the beauty of MMM is not on the results only, but on being able to drive changes in decisions, and there’s nothing more human that organizational change.
Industry updates and events
Planning your 2024 agenda? Have these two events in your radar:
I-COM Summit 2024, to be held in Málaga (Spain) in May. Registration
MeasureCamp Czechia, to be held in Prague (Czech Republic) in September. Registration
ARF’s “Attribution & Analytics Accelerator 2023” materials are available on their website (link), for members.
Chart of the day
Something really interesting has been happening in the fashion sector for some months. Really interesting industry to follow.
Learning corner
Is learning something new in your 2024 resolution list? The video series “Making Friends with Machine Learning” (YouTube playlist) is a great suggestion. Great overview of the process of running successful ML/AI projects, from a business and technical standpoint, delivered by the true expert Cassie Kozyrkov.
Incrementality testing is one of the skills that is becoming a must for every marketer. Review this course on incrementality experiments from Google. An hour of your life very well spent!