Attribution Windows Explained: Why One Setting Quietly Changes Every Client's ROAS

A client emails: "Our ROAS just dropped 30% — what happened?" You open the account, and performance is fine. Spend is steady, conversions are steady, the backend numbers in Shopify haven't moved. What changed was a setting most people never look at: the attribution window. It is the single most powerful dial in ad reporting, and almost nobody on the client side knows it exists.

Here is what the window actually does, why the numbers move when performance doesn't, and how to present it so it never triggers a fire drill.

What is an attribution window?

An attribution window is the period after someone clicks or views an ad during which a resulting purchase gets credited to that ad. Meta's long-standing default was 7-day click plus 1-day view — a sale counts if the person clicked within seven days, or saw the ad within one day and then converted.

That definition is doing a lot of quiet work. Widen the window and the same campaign "earns" more conversions, because you are crediting purchases that happened days after the click. Narrow it and those same sales fall off the report — even though the customer still bought. The window is not measuring performance; it is deciding how generously to assign credit (The HQ Digital).

How much can the window move ROAS?

A lot — 20–60% in either direction, with no change in real business outcomes. Moving from a 1-day click window to a 7-day click window typically lifts reported ROAS by 30–50% overnight.

This is the number every agency should have memorised, because it explains most "sudden" swings. Choosing the wrong window inflates or deflates reported ROAS by 20–60% without changing what the business actually earned (Get Ryze). Going the other direction — switching from 7-day click plus 1-day view to 7-day click only — cuts reported ROAS by 15–35%. That is not a performance drop; it is just more honest reporting (Analytigrow).

What changed with Meta attribution in 2026?

Two things, and both moved the numbers. On January 12, 2026, Meta permanently removed 7-day view and 28-day view windows from the Ads Insights API. In March 2026, it redefined what counts as a "click," pushing social interactions into a separate engaged-view category with only a 1-day window.

Together, those shifts caused reported conversions to fall 15–40% overnight across millions of accounts — not because campaigns got worse, but because the measurement methodology changed underneath everyone at once (DOJO AI). If a client's dashboard dropped a step this year and you couldn't find a cause in the account, this is almost certainly why. The lesson for reporting is blunt: a year-over-year ROAS comparison that straddles these dates is comparing two different rulebooks.

Does Google Ads have the same problem?

Yes, in its own form. Google shifted its default click window and shortened longer windows, and reported conversion volume tends to drop initially when delayed-conversion credit is removed — even when actual sales are flat.

When Google moved from a 30-day to a 7-day view, the shorter window revealed how much late-converting credit the longer window had been carrying (Search Engine Land). The takeaway is the same across platforms: the window is a reporting choice, and every platform makes that choice differently, which is a large part of why your Google, Meta and GA4 numbers never agree in the first place.

How should you report attribution windows to clients?

State the window on every report, keep it fixed, and reconcile against backend truth. The window should be a visible, boring line on the dashboard — not a hidden default that surprises everyone when a platform changes it.

Three practices keep this from becoming a monthly panic. First, label the window explicitly on every client report ("Meta: 7-day click") so a change is visible, not mysterious. Second, hold the window constant when comparing periods — if you must change it, restate the prior period on the new basis so you are comparing like with like. Third, sanity-check platform ROAS against backend numbers: if view-through conversions are more than about 25% of the total, the reported ROAS is probably flattered, and the honest figure is closer to the 7-day-click-only view (Get Ryze). Do this and the client stops asking "what happened," because the answer is already on the page.

FAQ

Which attribution window is "correct"? None of them is objectively correct — each answers a different question. 7-day click only is the more conservative, backend-aligned view; wider windows credit more assisted conversions. Pick one deliberately, document why, and stay consistent.

Why do Meta and my Shopify numbers disagree? Meta credits a sale to an ad if it falls inside the attribution window, using its own click and view definitions; Shopify counts orders. The gap is the assisted and view-through credit that Meta claims and your store's order log does not. Comparing the two is exactly how you find the honest number.

Should I use 1-day or 7-day click for Meta? Test both against your backend. If view-through and late conversions make up a large share, 7-day click can flatter you; 1-day click is stricter and usually closer to what the business banks. The right answer depends on the account, not on a universal rule.

Did my campaigns actually get worse in 2026? Probably not, if the drop lined up with January or March 2026. Those were methodology changes to how Meta counts, not changes in customer behaviour. Check backend revenue over the same window before touching budgets.

---

At Sifra we build agency reporting that puts the attribution window, the backend number and the platform number side by side — data, made visible — so a methodology change never lands as a client emergency. See our agency reporting work, or get a free mock dashboard built on one of your client accounts to see how it reads.