A lookback window is the trailing time horizon a system uses to decide which historical events count toward a current measurement or audience-building decision. The same control name shows up in two different surfaces of any ad manager — one governing conversion credit, the other governing audience eligibility — and operators routinely conflate them. The shape is identical (a trailing window applied to a backward-looking question) but the question is different in each case, so naming which kind of lookback is in play is usually the first step before tuning one.
Attribution lookback
The attribution lookback is the window before a conversion within which a click or view is eligible to receive credit. The click-versus-view split is the same idea applied to two different signals — a deliberate ad interaction versus a passive impression. Click-window defaults have historically sat in the 7- to 30-day range; view-window defaults are typically around a day; all are configurable, often per conversion action. Current defaults shift with platform releases, so the platform UI is the source of truth, not memory. For the deeper treatment of how the window shapes credit assignment, see attribution window, which is the specific case of a lookback governing attribution credit.
Audience lookback
The audience lookback is the window before “now” within which a user’s behavior makes them eligible for a remarketing or lookalike audience. This is a different decision: it governs who enters the audience, not who gets credit for a future sale. Most audience builders default in the 30- to 90-day range and are operator-tuneable up to platform-specific maxima of several months.
Consider an operator building a cart-abandoner retargeting audience. A 7-day window keeps intent dense — the user is still actively shopping — but the audience may be too small to feed the platform’s delivery algorithm. A 60-day window swells the audience but pulls in users who have moved on. A 30-day window is a common default because it balances the two. The tradeoff cuts the same way at every length: longer windows yield bigger audiences but lower intent density; shorter windows give tighter, fresher audiences but starve the algorithm of training signal.
The configured window is a ceiling, not a guarantee
The window you set in the UI is a ceiling on the effective window, not a guarantee. Privacy changes — iOS’s App Tracking Transparency, browser restrictions on third-party cookies — have shortened the de facto lookback for many use cases because the underlying user-identification signal expires faster than the configured window allows. The platform credits or includes only what it can still resolve; what it can’t resolve drops silently, so the configured window and the effective window can diverge without any visible signal in the report.
When you encounter a “lookback” control in any platform, name which decision it governs — attribution credit or audience eligibility — before tuning it. The right length is governed by different forces in each case.