Short answer
Marketing attribution is how a tool decides which marketing touch gets credit for a sale.
A real customer path looks like this: sees an Instagram ad on Monday, googles the brand on Wednesday, opens an email on Friday, buys. Four touches, one order. Attribution is the rulebook that splits (or doesn't split) the credit.
Why it matters
Every tool has its own rulebook, and each one is a little generous to itself:
- Ad platforms (Meta, Google, TikTok) count a sale if the buyer interacted with their ad within a time window, regardless of what else happened.
- Analytics tools (Google Analytics, Shopify reports) see the visit that led to the purchase and usually credit that.
- Email tools credit a sale to an email the customer opened days earlier.
So one order can show up as a win in three dashboards. That's why your numbers never match, and why "which channel works?" is harder to answer than it looks.
When attribution breaks (a tracking pixel stops firing, a checkout change drops UTM tags), spend decisions are made on bad data. The first symptom is usually a CAC that jumps for no obvious reason.
Common models
| Model | Who gets the credit | Good for |
|---|---|---|
| Last click | The final touch before purchase | Simple, but over-credits branded search and email |
| First click | The first touch | Seeing what introduces people to you |
| Linear | Split evenly across all touches | Seeing the whole path |
| Data-driven | A model estimates each touch's contribution | Large stores with enough data |
| View-through | Someone saw an ad and later bought, without clicking | Platforms like it. Treat with caution. |
An attribution window is the time limit for credit. A "7-day click" window means the ad gets credit if the click happened within 7 days before the purchase.
Example
Illustrative. The Monday-to-Friday customer above buys a $100 order.
- Meta reports $100 (ad click within its window).
- Google Ads reports $100 (brand search click).
- The email tool reports $100 (email opened before purchase).
- Shopify shows one $100 order.
Three tools, $300 of "attributed" revenue, $100 in the bank. Nobody lied. Everyone just used their own rules.
What to do about it
- Pick one source of truth for revenue: your store's orders, not ad dashboards.
- Use MER as the top-level check, because it doesn't depend on attribution at all.
- Use UTM tags consistently on every link you control, so analytics can see where visits come from.
- Test incrementality when spend is big: turn a channel off in one region or for two weeks and see what actually changes. It's the only way to know what a channel really adds.
Often confused with
- Tracking: collecting the data (pixels, UTMs, server events). Attribution is the rule applied to that data. Bad tracking breaks any model.
- Incrementality: what sales happened because of a channel, versus sales that would have happened anyway. Attribution guesses; incrementality tests measure.