adlaneSign in Metric note
Marketing attribution: definition, models and what credit means
Marketing attribution is the set of rules a measurement system uses to assign credit for an outcome, such as a purchase, to the interactions that preceded it. In formula terms, attributed conversions are the outcomes observed within a window, multiplied by the share of credit the model gives to each interaction. The result is an interpretation of a journey, not proof that the outcome would not have happened without that interaction.
By Adlane · Updated
What does an attribution model actually decide?
Three settings do most of the work. The window decides how long after an interaction an outcome can still be credited. The interaction type decides whether views count or only clicks. The model decides how credit is split when several interactions qualify, for example all to the last click or shared according to a data-driven estimate.
Google Ads sets these per conversion action and reports under the account’s chosen model; Meta sets an attribution setting per ad set and reports results under it. Record all three settings beside any attributed figure, together with the report date, because outcomes keep arriving until the window closes.
Why each platform reports its own credited outcomes
Illustrative example: a customer clicks a Google search ad on Monday, sees a Meta ad on Wednesday and buys once on Thursday. Under its rules, Google credits a purchase to the Monday click. Under its rules, Meta credits a purchase to the Wednesday view. The business recorded one order. Both reports are correct under their own definitions, and adding them gives two.
The table sets the three views side by side.
| Source (illustrative) | Credited purchases | Rule applied | Can be added to the others? |
|---|---|---|---|
| Google Ads | 1 | Click within window, account model | No |
| Meta Ads | 1 | View or click within ad set setting | No |
| Order system | 1 | Recorded order | This is the reconciliation figure |
Why changing a setting changes results without changing sales
Switching from a 30-day to a 7-day window, or from last click to a data-driven model, redistributes credit between campaigns and can raise or lower the total credited conversions. Sales did not change; the interpretation did. Compare only reports produced under the same settings, or explain the difference explicitly.
Cost per attributed conversion inherits all of this. It is also undefined when zero conversions are credited, and should then be shown as unavailable rather than zero.
Credited outcomes are not incremental outcomes
Attribution organises observed interactions; it does not test what would have happened without them. A branded search click credited with a purchase may have come from a customer who had already decided. Only an experiment with a control group, such as a geographic holdout or a platform lift study, estimates the incremental effect, and even those carry uncertainty.
A dashboard or AI summary that combines attribution reports cannot infer incrementality from them. When the decision requires a causal estimate, say so and seek stronger evidence before reallocating budget.
How to reconcile attribution claims with business records
Keep each platform’s credited figure with its settings, and keep the business’s recorded orders or qualified leads as a separate line. The gap between the sum of platform claims and recorded outcomes is a normal feature of attribution, not an error to be hidden by adjusting one side.
State inferences separately from observations. “Google credits 40 purchases, Meta credits 55, the shop recorded 70” is an observation; “Meta drove most of the growth” is an inference that needs more support.
How to review attribution in Adlane
Connect the Google Ads and Meta accounts for the business and ask a bounded question such as “For September, show attributed purchases and conversion value by campaign for each platform separately, with the attribution setting for each, in the account currency.” Adlane keeps the two sets of evidence apart and will not present a combined conversion total.
Any change you decide on afterwards, such as pausing a campaign whose credited results fell, becomes a proposal that you approve or reject. A free CPA calculator on the site can be used for the cost-per-outcome arithmetic once you have settled which outcome and setting you are using.