Monday morning, one account, one dropdown. I moved the ad sets to 1-day click because a client wanted conservative numbers, and by Wednesday the reported ROAS had fallen from 3.1x to 2.2x. I spent that Wednesday hunting for a delivery problem. There wasn't one. Orders were flat all week.

The bank account had not noticed anything happen.

Short answer: Meta's attribution window sets the rule for which conversions get credited to an ad, not how many purchases happened. Switch from 7-day click to 1-day click and the same ad reports less revenue. Two controls carry that name: the ad set setting, which also retrains delivery, and the Ads Manager reporting comparison, which changes nothing.

The takeaways

The default is 7-day click plus 1-day view, per Meta's documentation, so the ROAS you have been scaling on already counts purchases made up to a week after the click.
Two ROAS figures measured on different windows do not compare. A client's spreadsheet, an old screenshot and an agency benchmark are all unusable until someone names the window behind them.
Changing the setting on a live ad set is a campaign edit. Meta's documentation says significant ad set edits restart the learning phase, and at a €40 target CPA the 50 events it wants back are about €2,000 of conversion spend.

What does the attribution window change?

It changes the deadline for claiming credit. Under 7-day click, a purchase made within seven days of clicking the ad counts as that ad's result. Under 1-day view, a purchase made within a day of an impression nobody clicked counts too. Meta's documentation puts the account default at 7-day click plus 1-day view.

Nothing in that rule reaches your checkout. Shorten the window and Meta disqualifies the slower purchases it used to claim. The creative did identical work; fewer of its results now clear the deadline in time.

Which is why "ROAS" and "reported ROAS" should stop being synonyms in the same meeting. One is money that arrived. The other is money the platform will claim under a rule you picked from a dropdown.

Why did my ROAS drop the day I switched to 1-day click?

Because you narrowed the claim deadline and every conversion sitting outside it stopped counting toward that ad. The purchases still happened. They just stopped being the ad's purchases in the report.

How much falls out is specific to your account, and that is the honest limit here. It depends entirely on how long your buyers take. A €19 impulse product loses almost nothing, because those people buy in the same session. A €900 considered purchase loses a lot, because half its buyers come back on day four.

So treat any fixed percentage you read as a description of someone else's account. Run the check yourself: pull store orders for the same week, before and after the switch. If orders held flat while reported ROAS fell, you moved the measurement.

Which control did you move, the setting or the view?

Meta puts the attribution window in two places and gives both the same name, which is where most of the confusion lives.

In the ad set, it is a setting. It tells the delivery system which conversions count as the goal, so it shapes who Meta goes looking for. Moving it edits the campaign, and Meta's help center says significant ad set edits restart the learning phase.

In Ads Manager, under the column settings, it is a reporting comparison. It re-reads the same history under a different rule. It touches nothing live, costs nothing, and is the right place to answer "how would this look on 1-day click?" before you commit.

Guides describe both and call them one thing. Before anyone argues about a number, say out loud which one got touched. Check the labels as you go: Meta has moved these options before.

Can I compare two ROAS figures on different windows?

No, and this is the expensive one. A client's Q1 spreadsheet, a screenshot from last year, a benchmark in an agency deck, a case study on LinkedIn: every one of those was measured under some window, and almost none of them say which.

Two numbers produced under different rules describe different questions. Putting them side by side in a deck creates a trend that never happened, and I have watched an account get restructured on exactly that illusion.

The fix costs nothing. Write the window next to every ROAS figure you report: in the cell, in the caption, in the Slack message. A ROAS number with no stated window is an anecdote. One habit, and the quarterly review stops being an argument about screenshots.

Is a shorter window the safer choice?

Shorter windows are stricter. Strictness and accuracy are different properties, and it helps to see both ends fail.

At the long end, patient categories get undersold. A considered purchase that lands on day nine is invisible to a 7-day window no matter how good the ad was, so the channel looks weaker than it is and the budget goes elsewhere.

At the short end, the distortion is sneakier. A 1-day window flatters bottom-funnel retargeting, because those people were already mid-decision, and it starves prospecting, because cold buyers rarely decide inside 24 hours. Judge a whole portfolio on it for a quarter and your spend drifts toward the ads harvesting demand somebody else created.

Pick the window that matches how long your buyers take. Then leave it alone long enough to read a trend.

So which window is the true one?

None of them. Every window is a rule for assigning credit, and credit is not the sale. The only ledger that settles anything is your own checkout.

This is also why Adscalr does not set, change or reconcile your attribution window. It reads whatever the connected account reports. What it does inherit is the consequence. The composite score that ranks a creative is built on six metrics including ROAS, and the ROAS floor safeguard that pauses an ad above 1.5x rather than killing it reads the same number. Both are only as pinned down as the window underneath them. The rest of that thinking runs through our ad intelligence approach.

The window is also the dial that produces the gap between the platform's count and your store's, which I walked through in why Meta conversions don't match your sales. And if you run Meta against TikTok and Google at once, each platform defaults to its own window, so standardize that before you compare the three.

Meta attribution window: 7-day vs 1-day click