Man sieht nur, was man weiß” (Johann Wolfgang von Goethe) – One only sees what one know)
About the Author: Weiwei Liu-Schröder is Data & Measurement Lead at Google (Sponsor of the 2024 DDMA Digital Analytics Summit)
Working at Google, Goethe’s proverb served me as a reminder to invest time in learning the fundamentals before jumping into interpretation of data. One common behavior I have seen within marketing teams is that they first decide on a storyline and then scramble to find the data to match that storyline and justify ad spend. That is why I want to advocate for the role of analytics as a strong foundation before moving into any investment in attribution, incrementality tests or MMMs or even before paid campaigns. If you have a brand new website, think of establishing a baseline of your website’s current performance. This includes understanding your organic traffic, user behavior, and conversion rates. Having this baseline data provides a reference point to measure the impact of your paid investments later on. Some people take Analytics for granted, they even forget to update their tagging which can prevent data from being captured accurately. This poses the risks of not providing a reliable foundation and thus undermining the tagging infrastructure. To learn more about how to build a durable ads infrastructure with consented data, have a look here.
In the last years I heard many marketeers asking the questions:
- Which marketing channel drove most users to our website?
- Which users convert with the largest basket?
- Can we see which platform brings us the most revenue?
To answer these questions, my colleague Ana Carreira Vidal, together with a group of 40 Google experts, led the creation of a playbook around Modern Measurement this year which you can download here.
In this playbook she discusses different maturity stages for media effectiveness and proposes a framework. Interestingly she states that no tool has all the answers, but that most mature frameworks use multiple tools. She also states the need to invest in a durable measurement setup with privacy-centric measurement tools that maximize observed data and leverage first-party data. I want to expand that view to incorporate Analytics to gain a holistic understanding of Modern Measurement.
- Analytics – Foundational Pillar for Modern Measurement
Analytics is a non-negotiable tool that helps businesses track their website traffic and measure the effectiveness of their marketing campaigns. However, Analytics can also be used as the foundation for a modern measurement strategy that includes attribution, incrementality testing, and marketing mix modeling (MMM).
Gaining insights into user behavior is crucial to steer your online efforts. Therefore I see huge benefits of starting with Analytics to gain a holistic view and to start streamlining marketing efforts across different platforms. For example, by leveraging the advanced audience capabilities of Google Analytics, combined with the power of BigQuery and Looker, you can gain a deeper understanding of your customers and drive better business outcomes. Having worked in Google Analytics the last few years, I witnessed that the seamless integration with other Google products like Google Ads and the Google Marketing Platform helped unlock some of those first party insights.
How to use: Measure with analytics continuously, before moving onto other pillars.
- Attribution – Underestimated Pillar of Modern Measurement
Attribution is the process of assigning credit for conversions to different marketing channels or touchpoints. This information can then be used to optimize marketing campaigns and allocate budgets more effectively. It has historically been a very valued tool but also at the same time being questioned and underestimated in modern digital marketing. However, in a world where customer journeys have become increasingly complex, pitfalls in attribution, such as traditional models, fail to acknowledge the contribution of earlier touchpoints. Data is often siloed, making it difficult to understand the full user journey. This lack of unified data hinders accurate attribution and makes it difficult to determine which touchpoints played the most significant role in driving the conversion.
The notion that ‘attribution is dead’ has been circulating among marketers. However, a more accurate assessment is that attribution is undergoing a significant transformation to prioritize privacy and long-term sustainability. While our reliance on traditional conversion tracking might diminish over time, the combination of sophisticated modeling techniques and privacy-conscious measurement solutions will enable advertisers to continue effectively evaluating the impact of their ad campaigns. Data-driven attribution will remain the cornerstone of optimizing day-to-day ad activities, offering granular insights, incrementality calibration, and seamless integration with AI-powered bidding strategies.
To steer on value, your attribution model is crucial, as it might be the main source of data. Might it be tROAS or might it be tCPA, your attribution defines your conversion value.
How to use: Look at attribution regularly and on an ongoing basis.
- Incrementality Testing
Incrementality testing is a method for measuring the impact of a marketing campaign by comparing the results of the campaign to a control group. It uses randomized controlled
experiments to compare the change in consumer behavior between groups that are exposed or withheld from marketing activity while keeping all other factors constant. They are becoming more accessible and popular among advertisers, thanks to more open source resources and increased availability to run experiments on the platform. Analytics can be used to track campaign performance and measure the lift in conversions that is attributable to the campaign. It is the gold standard to measure causality, so it gives the most rigorous view of the incremental value brought by the marketing investment. It gives a snapshot of a concrete strategy at a concrete point in time.
How to use: Use Incrementality Testing every quarter.
- Marketing Mix Modeling (MMM)
MMM is a statistical technique that can be used to measure the impact of different marketing channels on sales. It uses top-level modeling that utilizes advanced statistics to understand
what drives sales. It measures media investment efficiency on top of base sales and other external factors that impact sales (e.g. seasonality, pricing, economy). It gives a holistic overview of all channels, sales, and external factors and can also provide a longer-term view of
media impact. It doesn’t require user-level data, making it more future-proof. Because it requires modeling with causal inference assumptions and at least two years of historical data, it can be expensive to run. Analytics can be used to provide data for MMM, such as website traffic, conversion rates, and revenue.
How to use: Twice a year (However, some advanced advertisers do it quarterly)
“千里之行，始于足下” (Laozi)
The journey of a thousand miles begins with a single step.
Born in China, I have been raised with the values of perseverance and persistence. That is why I want to close off with an emphasis on the importance of taking small, consistent actions towards achieving a larger goal. It is important in the world of data and measurement to break down the mountain of data into specific metrics and goals to derive actionable insights. It requires incremental progress by continuous analysis, experimentation, and refinement.
I recently had a discussion with a client who wanted to start on the topic of media effectiveness. When I asked them what their hypothesis was, they blanked. They actually didn’t think about what exactly they wanted to test. When I added the question of “What do you want to change if the hypothesis is proven right or wrong?”, a long pause told me that they didn’t really think about it before starting. As Ana stated in her playbook: “each experiment should aim to answer a business decision that is relevant enough to warrant designing an experiment”. Sometimes it might be better to perform a Causal Impact Analysis or a simple A/B Test. Sometimes the answers can be found in your existing data set within your analytics tool like Google Analytics.
If you are new to the Analytics world, there are multiple ways to get started. There is a large and active community of users, developers, and experts out there, starting from broader ones like the Measurecamp community to more specific ones like the Google Analytics community or local Women in Analytics groups, vibrant communities in the world provide a wealth of resources, including forums, blogs and tutorials making it easier to learn, troubleshoot, and stay up-to-date with the latest developments.
I encourage you to get the fundamentals right by building on a solid foundation of Analytics before moving on to combine more measurement effectiveness tools like attribution, incrementality and MMM.
Typically, most mature frameworks use all four pillars: Analytics, Attribution, Incrementality and MMM with Analytics being non-negotiable. Analytics is more than just tracking website traffic. It functions as the base for a modern measurement strategy, fueling the understanding of what marketers know, because in the end “one only sees what one knows”.
Weiwei Liu-Schröder
Ook interessant
Third-Party Cookies: Where We Are and What’s Next
Look-alike Modeling: A Comprehensive Guide to Customer Acquisition
