How AI Is Making Marketing Attribution Easier

Written by Scott Desgrosseilliers | Aug 31, 2026, 10:08:58 AM

How AI Is Making Marketing Attribution Easy: Scott on Ecommerce Coffee Break

I joined the Ecommerce Coffee Break podcast to talk through something that trips up a huge number of brands: why the attribution data they rely on is quietly leading them astray, and how to fix it.

The root problem is that most brands lean on ad platform reporting to tell them what is working. The trouble is that every platform has an incentive to make itself look good, so it claims credit generously and reports on its own performance. Act on those numbers and you end up over-investing in whatever channel is best at taking credit, rather than whatever is actually driving new customers.

A big part of the conversation was about lookback windows. A default 30-day window sounds reasonable, but it can create false attribution signals, crediting a channel for a conversion it did not really cause, or missing the genuine first touch entirely. The window you use has to match the buying cycle you are actually measuring, not a platform default. Get that wrong and your scoreboard is telling you a story that is not true.

From there I got into the fix, which is having a real measurement strategy instead of trusting the platforms. That means prioritizing new customer acquisition in your paid media, and being able to segment new customers from repeat ones, because if you cannot tell them apart, you cannot tell whether your ads are growing the business or just re-converting people who were already coming back. It is also the reasoning behind the Five Forces system, which is designed to reveal whether your marketing is actually aligned, and to distill all that data down to a clear call: scale, chill, or kill.

We also talked about where AI fits. Used the right way, with strict guardrails, AI can take a huge amount of the manual work out of attribution and make accurate measurement far more accessible. Used carelessly, it invents confident answers that are wrong. The difference is entirely in how tightly it is governed.

If you run an ecommerce brand and you have ever suspected your attribution is not telling you the truth, this one is a quick, practical listen.