Attribution Systems

Every platform claims credit for the same conversion, and none of them will tell you the truth. Meta, Google, TikTok and Snapchat each count a sale as theirs if their ad was anywhere near it, inside attribution windows they set themselves. Add their reports together and you can easily end up with more sales than you made.

I got tired of arguing with dashboards, so I built my own attribution.

What it does

• Tracks at the code level, on the site itself. Visits, steps and orders are recorded first-hand, rather than reconstructed from whatever each platform chooses to report.
• Follows one visitor across sessions and channels. Someone who sees an ad on one platform, searches on another and buys days later on a different device is one journey, not three unrelated clicks.
• Counts every order once. Credit is shared across the touchpoints that led to a sale, so the total adds up to the sales that actually happened.
• Keeps measurement and interpretation apart. Where a customer came from and where they stopped are recorded as facts. Why they stopped stays a hypothesis until it's tested.

What building it taught me

Most of the work is plumbing. Clean, connected event data matters more than any model built on top of it. If the plumbing is wrong, no algorithm will save you.

And the model is never the final word. The honest test of whether a channel works is to change it, cut its budget for a couple of weeks for example, and watch what happens to the total. Attribution tells you where to look. Experiments tell you what's true.

What it changed

It worked better than what I'd been using, and it changed how I think about the whole field. Once the numbers add up, the conversation moves from whose dashboard is right to which decision to make.

Read more

• Every Platform Took Credit for the Same Sale