Attribution answers one question: which marketing earned this sale? Here is how it works, where the common models go wrong, and what to look for.
The question behind the reports
A customer might see an Instagram ad on Monday, search for you on Wednesday, open an email on Friday and buy on Saturday. Four touches, one sale. Attribution is the work of connecting that sale to those touches, so you know which marketing to keep paying for.
The common models
- First touch gives all the credit to the first thing that brought the person in. It rewards the channels that create demand, such as social and video.
- Last touch gives all the credit to the final click before the sale. It rewards the channels that close, such as branded search and email.
- Last marketing touch is last touch with direct visits ignored, because typing your address in is rarely the reason someone bought.
- Multi-touch models share the credit across the path: evenly, more to recent touches, or by position.
No model is “true”. Each one is a lens. The useful habit is to look at two or three side by side: a channel that looks weak on last touch but strong on first touch is probably feeding the rest of your funnel.
Where attribution breaks
- The visitor is forgotten. Browsers and ad blockers cut cookies short, so a returning buyer looks like a new one and the first touch is lost.
- The conversion is invisible. A phone call, a chat or a deal closed in the CRM never reaches the analytics tag.
- Every platform counts itself. Google, Meta and Microsoft each claim the same sale, so platform totals add up to more than you sold.
- Tags are inconsistent. “fb”, “Facebook” and “facebook.com” end up as three different sources.
What good attribution looks like
- It starts from your real conversions (orders, qualified leads, booked calls) rather than platform-reported ones.
- It keeps each person’s whole path, so you can change the model without losing history.
- It includes calls, chats and offline sales.
- It cleans up campaign tags when reports are read, keeping the originals.
- It sends the result back to the ad platforms, so their bidding learns from real customers.
That is the approach behind Metric Sensei Attribution.