adlaneSign in Practical guide
Meta attribution window: read credited outcomes correctly
A Meta attribution window defines which ad interactions can receive credit for an outcome and how long after the interaction that credit can be given. Changing the window changes reported outcomes without changing the underlying sales. Record the setting with every comparison so readers know which question the report answers.
By Adlane · Updated
What a Meta attribution window actually decides
When someone buys after seeing or clicking an ad, Meta checks whether the purchase falls inside a rule: for example, within a number of days after a click or within a shorter period after a view. If it does, the purchase is credited to the ad under that rule. If it does not, the purchase still happened; it is simply not counted in that column.
That makes an attribution window a reporting rule, not a fact about the customer. Broader rules credit more outcomes; narrower rules credit fewer. Neither is the “true” number of sales, and both can be useful as long as the rule is stated. Treating the credited count as the sales count is where most misreadings begin.
Record the setting and the action before comparing two periods
Before comparing ratios between two periods or two campaigns, confirm both were read under the same attribution setting and that the same action is being counted. A change in default settings, a report rebuilt in a different tool or an export with a different window will all move the numbers while the campaign stands still.
Write the setting into the report header and into any question you ask about the data. Recently delivered ads have had less time for credit to accrue within the window, so the newest days will look weaker until they mature. Label them rather than excluding them silently.
Illustrative example: more credited purchases, same order count
Illustrative example: a fictional campaign is read under a narrow click-based setting and shows 40 credited purchases. The same campaign and period re-read under a broader setting that also credits views shows 64. The shop’s order system records the same number of orders in both cases, because nothing happened to the orders.
The 24 extra credited purchases are purchases that fell outside the narrow rule and inside the broad one. They are not evidence that the creative generated additional orders. A report that compares the 40 from last month with the 64 from this month, without saying the setting changed, would describe a 60 per cent improvement that did not occur. The figures are invented.
Separate delivery measures from attributed outcomes
Some measures depend on the attribution window and some do not. The table sorts common fields so you can see which comparisons stay valid when settings differ and which need the setting matched first.
| Measure | Depends on attribution window? | Comparable across differing settings? | Note |
|---|---|---|---|
| Spend, impressions, reach, frequency | No | Yes | Delivery is counted when it happens |
| Clicks of any type | No | Yes | The click is the event itself |
| Credited actions such as purchases or leads | Yes | Only with the setting matched | The rule decides what is counted |
| Cost per credited action | Yes | Only with the setting matched | Inherits the dependency from its denominator |
| Credited value or return on ad spend | Yes | Only with the setting matched | Also depends on currency and value implementation |
Two platforms can both claim the same purchase
A person can click a Google ad on Monday and a Meta ad on Wednesday, then buy on Thursday. Under their own rules both platforms may credit that order. Adding the two credited counts would record two sales where one happened. Platform claims are overlapping perspectives, not a single deduplicated ledger.
Reconcile against aggregate business records separately: the order system, the booking tool or the CRM. Keep each platform’s report intact with its setting attached, and treat differences as questions about rules rather than contradictions to resolve by picking whichever view looks strongest.
Keep the attribution window in the question you ask Adlane
In Adlane, connect Meta and select the ad account. Ask a bounded question that names the setting, such as “Compare this campaign’s credited purchases and cost per purchase for the last two complete months using the account’s current attribution setting, and tell me if the setting changed during that time.” Review the evidence with the setting shown beside the counts.
When settings differ between the periods you want to compare, explain the limit in the report rather than switching to the view that flatters performance. Any change that follows is prepared as a proposal you approve; nothing is applied from the conversation, and new campaigns start paused. Adlane never adds credited conversions across providers and is not affiliated with Meta.