Attribution windows explained: 24-hour vs 30-day vs lifetime cookies

A publisher sends a reader to a product page. The reader doesn't buy that day — they close the tab, think about it over the weekend, and come back three days later to check out. Whether that publisher gets paid for the sale they clearly caused depends on one setting most people never look at: the attribution window. Get the window wrong, on either side of the relationship, and money moves to the wrong place for reasons that have nothing to do with who actually earned it.

What an attribution window actually is

When someone clicks an affiliate link, the network needs a way to remember that click long enough to match it to a later conversion. Historically this was done with a browser cookie: the click drops a small file in the visitor's browser holding a unique identifier and a timestamp, and if a matching conversion event fires before that cookie expires, the click gets credit. The "attribution window" is simply how long that cookie — or, increasingly, a server-side click ID — stays valid before the opportunity to attribute a sale to it expires.

It sounds like a minor technical setting. In practice it's one of the highest-leverage numbers in the entire relationship between advertiser and publisher, because it directly decides how much of the real-world influence a piece of content has gets converted into an actual, paid commission.

There's also a distinction worth separating out early: a click-through window measures time from a click, while a view-through window — used far less often, mostly in display and video placements — measures time from an ad simply being seen, whether or not it was clicked at all. Affiliate marketing runs almost entirely on click-through attribution, since a link either gets clicked or it doesn't, but the underlying idea is the same in both cases: draw a line in time, and only count what happens before it.

The three common window lengths

Most programmes settle into one of three rough bands, each suited to a different kind of buying decision.

Short windows — 24 to 48 hours

Built for impulse purchases: low-cost consumer goods, flash sales, anything where the decision to buy happens in the same sitting as the click. A short window is cheap for the advertiser to run and rarely controversial, because most of the real conversions genuinely do happen inside that first day or two.

Medium windows — 7 to 30 days

The most common band across affiliate programmes generally, and the one that fits how most people actually shop: read a review, leave, compare a couple of alternatives, come back and buy within the week or two that follows. Thirty days has become something close to an informal default for exactly this reason — it's long enough to catch the realistic delay without stretching so far that credit starts going to clicks with little real bearing on the eventual sale.

Long and lifetime windows — 60+ days or uncapped

Used for considered, high-value, or subscription-style purchases — enterprise software, financial products, anything with a sales cycle measured in weeks rather than days. A "lifetime" cookie has no expiry at all and credits the referring click no matter how much later the conversion happens, which is generous to publishers but harder for advertisers to reconcile against a specific campaign period.

A worked example of a missed commission

Say a publisher sends a click on Monday. The visitor researches for three days and completes the purchase on Thursday — a completely normal, unremarkable buying pattern. Here's what happens to that commission under two different window lengths, on an otherwise identical $150 order at a 12% CPS rate.

Nothing about the traffic, the content, or the visitor's intent changed between those two scenarios. Only the window length did. This is the entire argument for taking attribution windows seriously rather than treating them as a fixed setting nobody revisits: the exact same referral can be worth full commission or worth nothing, purely as a function of a number set once during integration and rarely looked at again.

First-click vs last-click attribution

Window length answers "how long is the click valid." Attribution model answers a related but separate question: when more than one publisher's click falls inside a valid window before the same sale, who gets the credit?

Last-click attribution

The default on most networks, Kapvexa included. Whichever qualifying click happened most recently before the conversion gets full credit. It's simple to explain and simple to dispute, since there's exactly one click to point to.

First-click attribution

Credits the click that started the journey, on the reasoning that the content which first created interest deserves the reward even if a different publisher's link happened to be clicked right before checkout. Less common, and more complex to implement, since it requires holding onto the very first click ID across the entire window rather than just the most recent one.

Multi-touch and blended models

Some larger advertisers split credit — and the commission — across multiple qualifying touchpoints rather than awarding it all to one click. This is more equitable in theory but adds real reconciliation overhead, and most affiliate networks, Kapvexa included, default to last-click specifically because it's the model publishers can verify and trust without needing to audit an opaque split formula.

Why cookies alone keep getting less reliable

A cookie-based window only works if the cookie survives from click to conversion, and that assumption is breaking down. Browser privacy features increasingly restrict or auto-delete third-party cookies. Ad blockers strip tracking scripts before they can even set one. A visitor who clears their browser data, switches from their phone to a laptop, or opens a private browsing window between the click and the purchase breaks the chain regardless of how generous the window length is set to.

This is a different failure mode from the one covered in our guide to S2S postback vs pixel tracking, but the underlying lesson is the same: the more tracking depends on something happening correctly in the visitor's browser, the more conversions quietly go unattributed for reasons that have nothing to do with whether the publisher actually earned the sale.

Matching window length to payout model

The right window length isn't a universal number — it should track how long your specific funnel actually takes to close, which in turn tracks the payout model you're running. Our guide to CPS, CPL, and CPI payout models covers the funnel-stage differences in depth, but the attribution implication is straightforward: a CPI campaign paying for an app install that happens within minutes of the click needs only a short window, while a CPL campaign in insurance or B2B software — where the lead might not convert to a qualified opportunity for weeks — needs a window long enough to still be open when that happens.

Setting a short window on a long sales cycle doesn't reduce cost, it just moves real conversions into the "unattributed" bucket, where they still cost the advertiser the sale but no longer cost them the commission — which is a bad trade for the publisher relationship even though it looks like a win on a single line item.

It's worth running this check the other direction too: if you're a publisher deciding which campaigns to prioritize, a short window on a long-consideration product is a real signal about how that programme is likely to treat your reported earnings. Ask directly, before sending meaningful volume, rather than discovering the mismatch three weeks in when a batch of sales you know you influenced simply never shows up in the dashboard.

How window length distorts EPC

Every conversion that falls outside the attribution window is a conversion that never shows up in a publisher's earnings — which means it never shows up in their EPC calculation either. A campaign with a genuinely strong conversion rate can read as mediocre purely because its window is too short to catch the realistic delay between click and purchase for that particular offer.

This matters most when comparing EPC across campaigns with different window lengths, since the comparison isn't just measuring traffic quality anymore — it's partly measuring how generous each campaign's tracking setup happens to be. A publisher optimizing purely on reported EPC without checking window length can end up steering traffic away from a genuinely well-converting offer simply because its window is quietly discarding sales that happened a few days later than the cutoff.

Choosing a window as an advertiser

The trade-off runs in both directions. Set the window too short and you underpay publishers for sales their traffic genuinely caused, which erodes trust and eventually traffic volume as publishers notice their reported numbers don't match what they know they sent. Set it too long and you risk crediting a sale to a click that had little real influence — someone who clicked a coupon link weeks ago and would have bought anyway, now attributed entirely to that publisher.

A reasonable starting point is to look at your own average time-to-purchase data, if you have it, and set the window a comfortable margin beyond that — long enough to catch the realistic tail of normal buying behaviour without stretching so far that it starts rewarding coincidence over influence.

It also helps to treat the window as something you revisit rather than something you set once during integration and forget. Buying behaviour shifts with seasonality, with new traffic sources, and with changes to your own checkout flow — a window that was correctly sized a year ago may no longer match how your funnel actually behaves today, and the only way to know is to look at the data periodically rather than assume the original setting still holds.

Mistakes to avoid

FAQ