Meta Ads Attribution Window: Fix One Window per Creative Test
A Meta ads attribution window is set per ad set, so ads in different ad sets can be judged differently. Fix one window per test and compare every ad under it.
By Sachit Sharma, CEO & Founder · Updated 5 Nov 2026

Key takeaways
- Set the same attribution setting on every ad set in the test before it starts, and do not change it mid-test. A window change mid-test is a new test, not a reporting tweak.
- Meta sets attribution at the ad set level, so ads in different ad sets can be counted by different rules. Put the test ads in one ad set where you can.
- Before calling a winner, rank the ads under 1-day click and under 7-day click. If the leader changes, you do not have a winner yet.
- Read creative verdicts on click columns. View conversions count an ad that nobody interacted with, so they say little about which creative persuaded.
In this article
- 1What is the Meta ads attribution window and what does each setting count?
- 2How can the attribution window crown the wrong ad in a creative test?
- 3Which attribution window should you fix for a creative test?
- 4How do you check whether the window changes your winner in Ads Manager?
- 5What goes on a One Window Per Test card?
- 6How do the 2026 attribution changes affect comparing tests run before and after them?
- 7How does a fixed-window test card connect to our weekly test plans?
- 8Conclusion
- 9Frequently asked questions
Ads Manager will happily rank two ads that were never measured the same way. Each ad set carries its own attribution setting, and by default the results you see follow the ad set's setting, not the ad (Meta Business Help Center). Reports put ads from different ad sets side by side in one table, so the mismatch is easy to miss.
That matters most when you test creative. If ad A sits in an ad set that credits views and seven days of clicks, and ad B sits in one that credits one day of clicks, A can look better because its window is wider. The ad did nothing different. Its ruler was longer.
This page gives you a rule, a worked example with numbers, a three-column check to run in Ads Manager, and a test card you can copy. It gives a default pair of windows to read, 7-day click and 1-day click, and a rule for when the two disagree.
What is the Meta ads attribution window and what does each setting count?
The attribution window is the period in which Meta can credit a conversion, such as a purchase, to your ad. Ads Manager labels the control Attribution setting. It has three interaction types: click-through, engage-through and view-through, each with its own length.
Meta defines them in its attribution settings help article:
- Click-through counts events within 1 or 7 days after a link click on your ad.
- Engage-through counts events within 1 day after a non-link click action. That is any click on the ad except a link click. For video ads it also includes a 5-second play, or 97% of the video if it is shorter than 5 seconds.
- View-through counts events within 1 day after an impression of your ad, with no interaction needed.
That guide also reports that engage-through and view-through can each be turned off.
For a website conversion ad set, Meta's default setting in 2026 is 7-day click, 1-day engage-through and 1-day view-through, with every conversion counted. Click-through can be set to 1 day or 7 days. Engage-through and view-through run for 1 day only. The setting on an ad set decides both which conversions Ads Manager shows by default and how Meta optimizes delivery. (Jon Loomer Digital)
The windows overlap. A 7-day click window includes everything in the 1-day click window, so the columns must not be added together (Jon Loomer Digital, updated 1 March 2025).
How can the attribution window crown the wrong ad in a creative test?
The window can crown the wrong ad because a wider window produces more conversions for reasons unrelated to creative. Meta says results cannot be compared in the Campaign Overview table across ad sets with different attribution models and points to the Compare Attribution Settings feature (Meta Business Help Center). Meta's note names attribution models. The same logic applies to different windows, which is our inference.
The setting also changes delivery, not only reporting. It changes "both how your ad set is optimized for delivery… and how conversions are reported" (Jon Loomer Digital, updated 1 March 2025). Two ad sets with different settings are therefore different experiments.
Here is a worked example. The numbers are invented to show the mechanism, and the cost per purchase is spend divided by purchases. Both ads spent $2,000.

| Ad | Ad set setting | Default column | 1-day click column | 7-day click column |
|---|---|---|---|---|
| A | 7-day click + 1-day view | 40 purchases ($50.00) | 28 ($71.43) | 34 ($58.82) |
| B | 1-day click | 33 purchases ($60.61) | 33 ($60.61) | 37 ($54.05) |
In the default column, A wins by $10.61 per purchase. Under 1-day click and under 7-day click, B wins. A's lead came from view credit and its wider window, not from its creative.
Windows do move the numbers in real accounts, though modestly in this one. In a January 2026 example, one account showed 1,595 conversions under 1-day click and 1,668 under 7-day click, so 73 conversions, about 4% of the 7-day figure, fell on days 2 to 7 (Jon Loomer Digital, 19 January 2026). That is one account. Your gap between windows will differ, which is why you measure it.
Which attribution window should you fix for a creative test?
Set one click window on every ad set in the test, and give all of them the same one. We recommend 7-day click, because it leaves room for buyers who do not purchase on the day they click. If most of your buyers decide within a day, set 1-day click instead. Then read every ad under both 1-day click and 7-day click before you call a winner: the setting keeps the test fair, and the second read checks that the result does not depend on the window you picked. This is our recommendation.
The reasoning:
- Click windows tie a result to someone who acted on the ad. View-through credits an impression with no interaction, so it cannot tell you which creative persuaded anyone. Keep it out of the verdict, and look at it separately.
- The choice matters less than the consistency. Either click window gives a fair comparison if every ad has it.
- One ad set beats many. Ads in the same ad set share one setting by construction, because the setting lives at the ad set level. Put the creatives you are comparing together, and set the window when you build the ad set.
- Equal spend helps. Compare ads that have spent a similar amount, so a gap in results is not just a gap in budget.
If your buyers take longer than a week, a click window cannot see the whole path. In a ThoughtMetric analysis of 100 e-commerce brands (October 2025), the average time from first touch to order was 8 days, one day longer than Meta's 7-day click window. That figure counts first touch from any channel, not only a Meta click, so use it as a prompt rather than a rule: check your own store data for the gap between first visit and order. If most orders land after day 7, no ad set setting will credit them. Keep the same setting for every ad in the test anyway, and read the result as a ranking of ads, not a count of what Meta drove.
If the test must span several ad sets, give them the same setting and still read the verdict from the Compare Attribution Settings columns below. Our guide to analyzing Facebook ads and choosing your next test makes the same rule for attribution setting and reporting date range.
How do you check whether the window changes your winner in Ads Manager?
Add the comparison columns at the ad level and rank the ads under each window. Compare Attribution Settings shows results under other windows without changing delivery (Jon Loomer Digital).

- Set the date range to the whole test.Pass: every ad's range starts at its launch and ends on the same day.
- Open the Columns menu and select Compare Attribution Settings.
Tick 1-day click, 7-day click, 1-day engage-through and 1-day view. 28-day click is also offered there for reporting, but it is not an ad set setting you can optimize on (Jon Loomer Digital).
- Write down results and cost per result for each ad under 1-day click and under 7-day click.Do not add columns together. Pass: you have two figures per ad.
- Rank the ads under each window.Pass: the same ad leads under both. Fail: the leader changes, so you have no winner. Extend the test or rerun it.
- Remove the 1-day view column from your reasoning.Pass: the leading ad still leads without any view credit. Fail: its lead exists only with views included, so treat it as unproven. Then add the 1-day engage-through column to each ad's click result. Since 3 March 2026, saves, shares and other non-link interactions are credited there, not under click-through, so an ad that earns engagement can look weaker on click columns alone. Pass: the leader is the same with and without engage-through. Fail: the leader changes, so you have no winner yet; extend the test or rerun it. This is our recommendation.
A pass means the window did not decide the result. It does not prove the result is large enough to act on. For which metric answers which question, see our guide to Facebook ads metrics.
What goes on a One Window Per Test card?
The card records the window before the test starts, so nobody can change it unnoticed afterwards. Copy these fields into your test log.
| Field | What to write | Pass or fail |
|---|---|---|
| Test ID and hypothesis | One sentence naming the single variable | Pass: only one thing differs between ads |
| Attribution model | Standard or Incremental | Pass: the same in every ad set in the test |
| Window on the ad set(s) | For example 7-day click, 1-day engage-through, 1-day view | Pass: identical in every ad set |
| Date the window was set | Calendar date | Pass: before the first ad delivered |
| Windows read | 1-day click and 7-day click | Pass: written before results exist |
| Ads compared | Names, spend, results per ad under both windows | Pass: spend is similar across ads |
| Rank under each window | Ad order under 1-day click, then 7-day click | Pass: order matches |
| Window changed mid-test? | Yes or no, with date | Pass: no. If yes, the test restarts |
| Verdict | Winner, no winner, or rerun | If the orders differ, the verdict is no winner. Allowed only if every pass above holds |
Never mix Standard and Incremental ad sets inside one test, because the two models count conversions differently.
Put the window in the ad set name so the card and the account agree. Our Meta ads naming convention sets the ad set name as audience, segment, placement and event; add a short window code at the end, such as 7dc for 7-day click.
How do the 2026 attribution changes affect comparing tests run before and after them?
Do not compare results from before 3 March 2026 with results from after it, because click-through counted different things on each side. That is our rule. If someone changes an ad set's window during a live test, record the date on the card and restart.
Two dated boundaries matter, in order:
- 12 January 2026: Meta's Ads Insights API (the interface that reporting tools use to pull your data) stopped returning 7-day view and 28-day view figures (Meta Developers Blog, 16 October 2025). If a dashboard feeds your test card, those columns are not available through it.
- 3 March 2026: click-through narrowed to link clicks, and other clicks moved to engage-through (Jon Loomer Digital). Meta notes that some accounts may still use prior versions of click-through and engage-through while this rolls out (Meta Business Help Center). Confirm every ad set in your test uses the same definitions.
Do not read a drop in click-through conversions across 3 March 2026 as a drop in creative performance. The count changed; the ads may not have.
How does a fixed-window test card connect to our weekly test plans?
We combine ad-performance data, competitor activity, customer signals and past ad learnings into ranked weekly creative test plans and execution-ready creative briefs, which is what we mean by ad creative intelligence.
Ads Manager is where you set the window and read the columns. We work on the next decision: what to test, in what order, and why. A test card with the window written on it tells you whether last quarter's winner is evidence or an artifact of how it was counted.
To check whether an ad set has enough budget to settle before you judge it, read our guide to the Meta ads learning phase.
Conclusion
Before your next creative test, decide the window, not the winner. Give every ad set the same attribution setting, write it on the test card, and leave it alone until the test ends. When results arrive, rank the ads under 1-day click and 7-day click, then check the 1-day engage-through column. If the same ad leads each time, you have evidence. If the leader changes, you have a question to rerun, not a result to scale. And if your buyers usually take longer than a week to order, read the verdict as a ranking of ads, not a count of what Meta drove.
See how Deepsolv helps you understand what's working, why it's working, and what to test next. Book a demo to see it on your own test history.
See what your past tests really prove
Frequently asked questions
Sources
- 1.About attribution models and attribution settings, Meta Business Help Center
- 2.About actions attributed to your ad, Meta Business Help Center
- 3.Ads Insights API metric availability updates, Meta Developers Blog, 16 October 2025
- 4.How Meta Ads Attribution Works in 2026, Jon Loomer Digital, 10 March 2026
- 5.Attribution Breakdowns vs. Compare Attribution Settings, Jon Loomer Digital, 19 January 2026, updated 4 April 2026
- 6.Compare Attribution Settings: Get the Most of Meta Conversion Data, Jon Loomer Digital, 3 October 2023, updated 1 March 2025
- 7.Meta Ads Attribution Setting: A Complete Guide, Jon Loomer Digital, 12 June 2023, updated 1 March 2025



