Most customers don't buy after a single interaction. They might see a social ad, search for the brand a week later, open a newsletter and then order. Attribution is the set of rules that decides how much of that order each step gets credit for. The rules are called an attribution model.
The model you use changes the numbers you see. Under one model a paid social campaign looks unprofitable, under another it looks essential. That's why attribution sits behind almost every channel metric, including ROAS and cost per acquisition. You'll run into it in GA4, in the conversion settings of Google Ads and Meta, and in CRM tools that trace a closed deal back to the first form a lead filled out.
Mechanics
How attribution models split the credit
Rule based models apply a fixed formula. Last click gives all credit to the final touchpoint before the conversion. First click gives it all to the step that introduced the customer. Linear splits it evenly across every step. Time decay gives more weight to steps closer to the purchase, and position based models favor the first and last steps while sharing a smaller portion with the middle.
Data-driven attribution, the default in GA4 and Google Ads, compares paths that converted with paths that did not and estimates how much each touchpoint changed the odds. It only works well with enough conversion volume. Every model depends on tracking: without consistent UTM parameters, working pixels and server side signals, steps disappear from the path and their credit moves to whatever was recorded instead. And each ad platform reports conversions from its own view of the path, so the platform totals often add up to more than you actually sold.
Example
A worked example with three touchpoints
Assume a furniture store records 100 orders in a month, each with revenue of 600, for total revenue of 60,000. Suppose every order followed the same path: a Meta ad click, then a Google search ad click, then an email click.
Under last click, email gets all 60,000 of credited revenue and the two ad channels get nothing. Under first click, Meta gets all 60,000. Under linear, each channel gets one third, so 20,000 each. Now assume Meta spend was 8,000 and Google spend was 6,000. With linear credit, Meta ROAS is 20,000 / 8,000 = 2.5 and Google ROAS is 20,000 / 6,000 = 3.3. With last click, both show zero and look like the first thing to cut, even though the same customers would likely never have reached the email list without them.
Use
When attribution helps and when it misleads
- Use one agreed model across your reports, so month to month changes reflect performance rather than a switch in the counting rules.
- Compare two models side by side before moving budget, since a channel that looks weak under last click may be doing the introducing.
- Platform reported conversions mislead when you add them together, because Google Ads and Meta can both claim the same order.
- Attribution shows correlation along a recorded path, not proof that a channel caused the sale; holdout tests answer that question more directly.
Watch out
Common mistakes with attribution
- Cutting upper funnel campaigns because last click reports give them no credit, then watching branded search and direct sales slowly decline.
- Leaving links untagged in emails and social posts, so those visits land in direct or referral and their credit goes elsewhere.
- Switching models in the middle of a quarter and comparing the new numbers with the old ones as if nothing changed.
- Trusting data-driven models on accounts with very few conversions, where the estimates swing widely from week to week.
- Ignoring offline steps such as phone calls and sales meetings, which never appear in a web based path.
FAQ
Questions about Attribution
Which attribution model should a small business use?
Pick a model you can explain and keep it steady so month over month comparisons stay fair. Data driven attribution in GA4 works when you have a decent volume of conversions. With low volume, many owners use last click for daily decisions and glance at first click to see which channels introduce new buyers.
Why do Google Ads and Meta report more conversions than my store actually had?
Each ad platform credits itself whenever its ad touched the path, within its own lookback window. A shopper who clicked a Meta ad and later a search ad can be claimed by both. Treat platform totals as overlapping views and check them against the orders or leads recorded in your own system.
What is the difference between attribution and incrementality?
Attribution splits credit among touchpoints that were present before a sale. Incrementality asks whether the sale would have happened anyway without a given channel, usually answered with holdout or geo experiments. A channel can collect plenty of credit and still add little new revenue.
