Attribution window

The attribution window is the network's answer to "how far back are we willing to look for the click that caused this conversion?" In browser-only cookie tracking, the attribution window equals the cookie window. In server-side tracking with logged-in user matching, the attribution window can extend further because the network can still match a logged-in user to their previous click even after the browser cookie expires.

Cookie window versus attribution window in detail. The cookie window is a property of the browser: how long the cookie lives and is honoured. The attribution window is a property of the network's reporting logic: how far back into click history the network is willing to look when matching a conversion. For an anonymous browser, the two are identical. For a logged-in user or a click-ID handshake, the attribution window can extend well beyond the cookie window because the network has other ways to match the user.

Amazon Associates illustrates the distinction. The cookie window is 24 hours: if the buyer is browsing as a guest, only purchases within 24 hours of the click get credited. The attribution window is effectively longer for logged-in Amazon customers because Amazon can match the user's account history beyond the cookie. The exact behaviour is opaque (Amazon does not publish the rules), but creators consistently report attribution beyond the 24-hour mark for repeat shoppers and Prime members.

Per-network attribution windows. Awin runs 30 days default per merchant, with some merchants extending to 60 or 90. Impact lets each brand configure its attribution window between 1 and 720 days. CJ defaults to 7 to 30 days depending on merchant. ShareASale runs 30 to 90. Direct SaaS programmes commonly use 30 to 60 days; some go to 90 or even lifetime. The variance is enormous, and the spec is per-merchant rather than per-network.

Server-side identity matching. Three mechanisms extend attribution beyond the browser cookie. Logged-in user matching (Amazon, Booking.com, large platforms that recognise the visitor by account). Click-ID postbacks (the network keeps the click record server-side and matches the conversion via a server-to-server callback). Cross-device probabilistic matching (less common, more controversial; some networks attempt it via fingerprinting). All three sidestep the browser cookie and can preserve attribution well beyond the cookie's technical lifespan.

Holdback periods are not the same as attribution windows. Confusing because they look similar. The attribution window is "how far back do we look for the click?". The holdback period is "how long after the conversion do we wait before paying out?". A 30-day holdback exists so the merchant can reverse the commission if the buyer returns the product or cancels the order. Networks publish both numbers separately; do not collapse them.

Multi-device attribution. Buyers research on mobile and complete on desktop more often than networks would prefer. Most browser-cookie-only networks lose this attribution entirely (different device = different cookie). Networks with logged-in user matching (Amazon, Booking.com) preserve it. Direct-brand SaaS programmes that require account creation preserve it. For affiliate creators, this is one of the structural advantages of direct-brand programmes over Amazon Associates for high-consideration purchases where cross-device research is the norm.