What is an Attribution Window?

An attribution window (also called a lookback window) is the period after an ad click or impression during which a later conversion can still be credited to that touch. It is not an attribution model and not an answer to “who caused the purchase.” It is the shelf life of a touch: the click or impression is either still eligible under the crediting rules, or it has expired.

Product UIs use both names for the same setting. Attribution window and lookback window are synonyms. AppsFlyer and Adjust usually say lookback; Meta, Google Ads, TikTok, and web analytics say attribution or conversion window. There is no separate “lookback” concept — only platform jargon.

A model such as last-click decides who gets credit among touches that are still alive. The window decides which touches enter that set at all. Without a window, attribution is undefined: any click in a user’s history could claim a conversion months later.

Click window vs view window

Almost every system keeps two independent windows.

Click-through (click, CTW) — after a click. Typical values: 1, 7, 28, or 30 days. Search and CPA networks rely on this almost exclusively.
View-through (view, VTW, post-view) — after an impression with no click. Usually shorter: 1 day, sometimes 7. This is where a view-through conversion is born.

Priority is standard: if a click exists inside the click window, the view does not compete. Post-view fires only when there is no click or the click has already aged out. That is why “conversions in the ad account” and “paid conversions in the affiliate network” diverge even with clean tracking: the network pays on last click and cookie; the account can still add impressions.

Typical platform defaults

The figures below are common factory defaults, not a statute. Accounts let you compare windows and change the conversion action.

Meta Ads. After ATT the default moved toward a 7-day click and 1-day view (28/1 was common before). Reports can still isolate 1-day click.
Google Ads. Many website conversion actions default to a 30-day click-through window (often configurable 1–90) and a 1-day view-through window (up to 30) on display and video. Search campaigns live on clicks; view barely matters on Search.
TikTok. Often a 7-day click and 1-day view.
MMPs / apps. Classic setup: 7-day click lookback, 24-hour view lookback; click caps are usually 30 days. The SKAdNetwork window is Apple’s, not the buyer’s.
Affiliate networks. Cookie duration (7–30–90 days) is a cousin of the window: it decides who gets paid, not how ROAS is drawn in the ad UI.

How the window moves ROI and ROAS

A longer window “finds” more conversions for the source, raises reported ROAS, and lowers visible CPA. A shorter window is stricter: some late orders fall to organic, direct, or another campaign. The business did not change — only the crediting rule did.

Example. A user clicks a retargeting ad and buys 12 days later. Under a 7-day click window that purchase does not belong to the click; under 30 days it does. Retargeting ROI in the two reports differs with the same spend and the same orders. Reconciling tracker, ad account, and CPA network without aligning windows is pointless: it is three clocks on different time zones.

A long view window inflates warm channels in particular. Display and video with a 7-day VTW credit people who were going to buy anyway. Teams that run money on clicks often lock a short view (1 day) for operational ROI, or keep post-view out of the optimization goal.

Window and model are orthogonal

Last-click inside a 1-day window and last-click inside a 30-day window are the same model with a different candidate set. First-click and data-driven models use the same window as a history filter: only touches inside the lookback enter the path. Multi-touch without a window does not exist.

Vertical sets a sane horizon. Impulse offers and in-app purchases often close in hours or a day — 1–7 days of click is enough; a long view mostly adds VTC. Insurance, mortgages, and B2B lead gen stretch for weeks: 7 days will cut the top of the funnel. Subscription and gambling read FTD in a short window and count LTV separately, outside the ad lookback.

Discrepancies and privacy

Mismatched windows are the first reason for a report like “140 purchases in Meta, 90 in the tracker, 70 in the network.” Second is the model (network last click vs account data-driven). Third is view-through and modeled conversions. Fix one operating window (for example 7c/1v) and reconcile only inside it.

ITP, ATT, and the end of third-party cookies shorten the window you can actually realize: the UI may say 28 days while the user identifier dies earlier. Part of the platform report is then modeled. For a buyer that means comparing click conversions on one horizon — not the account’s long numbers against the network’s short last-click.