Stefan
Latest posts by Stefan (see all)
My first years as a Product Manager I worked on marketing analytics products, most notably Google Analytics. The measurement techniques we looked at for implementation were so exciting. Having studies Data Engineering & Statistics, I was always curious about the underlying math of what we did. So I forged strong relationships with the research and stats teams.
One of my biggest launches was the initial version of Data-driven Attribution in Google Analytics. I loved to show people the report, and explain how the model works. It was fascinating that we brought a concept we derived into our product. Along the way, product specific decisions had to be made, but we strived to stay true to the math behind it.
Attribution Modeling is a set of tactics to help you measure the effectiveness of your digital advertising.
A very smart and kind person
Now I don’t work on marketing analytics anymore, but I think about it often. I drafted countless documents which aimed to explain the models and the math behind it. The concepts behind an Attribution model – believe it or now – are actually not that complicated. It is the implementation and application of these concepts which takes years.
In this whitepaper you’ll learn what Attribution Modeling is, the basics of the models, and how each model calculates credit. The scope is limited to rules-based models. They are relevant to this day, since models such as Last Click still measure most of the worlds digital ad spend.
The paper covers fundamentals of Attribution Models, introduces the concept of ‘path data’ and rules-based models. It shows how attributed Return on Ad Spend (aROAS) is calculated for each model including formulas. It also touches on the complexity introduced by the rise in mobile devices, and shows how models can be applied in basic marketing decision making.
This document is only using information which is publicly available and does not contain information specific to any product implementation, including Google Ads products.
