Attribution models explained: what affiliates get paid on

An attribution model is the rule that decides which ad or touchpoint gets credit for a conversion. The main attribution models are first click, last click, linear, time decay, position-based and data-driven. For affiliates the model matters twice: once in the ad platform that reports results, and again in the program that pays commission. The two rarely agree.

What is an attribution model?

An attribution model takes every recorded touchpoint on the way to a conversion and decides how much credit each one gets. A touchpoint is any tracked interaction: an ad impression, a click, a visit from an email or a review page. Marketing attribution models differ only in how they split that credit.

Take one bettor in New Jersey. On Monday she clicks a sportsbook video ad that an affiliate runs on Instagram. It sends her to the affiliate's own landing page, not straight to the operator, and she leaves. On Wednesday she searches the sportsbook's name on Google and clicks the operator's own brand-search ad. On Friday she reads a comparison page on the affiliate's site, clicks the tracking link, registers and makes a first deposit. Three touchpoints, one deposit. Which one earned it?

The answer depends on who is counting. Each ad platform sees only its own touchpoints, your tracker sees the clicks you tagged with UTM parameters, and the operator's affiliate system sees the affiliate tracking links. Each applies its own model to its own slice of the journey.

The main attribution models

There are three groups of attribution models: single-touch, multi-touch and data-driven. Here is how each would credit the deposit in the example above.

Single-touch: last click and first click

Last click gives all the credit to the final click before the conversion. It is simple, easy to audit and close to how most affiliate programs pay, which makes it a good fit for short funnels with one channel. The weakness: every earlier touch counts for nothing, so prospecting ads look worthless.

First click does the opposite and credits the touch that started the journey. It helps when you want to know which channel brings new people in. It undervalues whatever actually closed the sale.

Multi-touch: linear, time decay and position-based

Multi-touch models spread credit across the path, which suits journeys with several channels and conversions that take days.

• Linear gives every touchpoint an equal share. Fair on paper, but a forgotten banner counts as much as the click that converted.

• Time decay gives more credit to touches closer to the conversion. It suits longer sales cycles where recent contact matters most.

• Position-based (U-shaped) gives most credit to the first and last touch and shares the rest across the middle. A 40/20/40 split is the common version.

Some analytics tools also let you build custom attribution models with your own weighting, which only works with clean data.

Data-driven attribution

Data-driven attribution learns the split from observed paths, comparing journeys that converted with those that did not. It adapts to your traffic but needs enough conversion volume, and it runs inside the ad platform, so you cannot audit the weights. Google Ads now uses it as the default.

Attribution windows: click vs view

An attribution window is how long after an interaction a conversion can still be credited to it. The window matters as much as the model, because the same campaign looks very different under a 1-day and a 7-day setting.

• Click-through window: the conversion happens within a set number of days after someone clicked the ad.

• View-through window: the person saw the ad, did not click, and converted later anyway.

• Lookback window: the analytics term for how far back a tool looks for touchpoints before a conversion.

View-through is where platform numbers inflate fastest. Someone who saw your retargeting ad may already have been on their way to the site, and a long view window still gives the campaign the sale. Before you scale a retargeting ad set, break its conversions into click and view.

Your own tracker usually records clicks only and attaches conversions through postback tracking, so it has no view-through credit at all. That alone creates a gap against platform dashboards.

How Meta and Google attribute conversions

The ad platforms report conversions under their own models and windows, and both have changed their options recently. We checked the settings below against each platform's help pages on October 29, 2026; check the current documentation before relying on them.

Meta (Facebook and Instagram)

Meta sets attribution at the ad set level. Its help center lists two attribution models, Standard and Incremental, plus a custom option that feeds attribution data from your own analytics tool:

• Standard credits conversions that fall inside the windows you choose.

• Incremental is Meta's setting that aims to credit conversions more likely to have been caused by the ad. In its incremental attribution page, Meta says machine learning models predict whether the ad caused the conversion.

Under Standard, the attribution windows for website conversions are:

• Click-through: 1 or 7 days after a link click.

• Engage-through: 1 day after any other click on the ad, meaning anything except a link click. For video it also includes a 5-second play, or 97% of a video shorter than 5 seconds.

• View-through: 1 day after an impression.

Click-through now counts only link clicks; other clicks sit under engage-through, though Meta notes some accounts may still run the earlier versions during rollout. The widest combination is 7-day click, 1-day engage-through and 1-day view-through. Meta warns that results from ad sets with different attribution models cannot be compared in the Campaign Overview table.

If you pull Meta reports by API, note a second change: since January 12, 2026 the Ads Insights API no longer returns 7-day view and 28-day view windows. Reports that still ask for those windows get no data for them. If you send server-side events, see how Meta Conversions API handles deduplication, since duplicate events also distort what Meta reports.

Google Ads and GA4

Google narrowed its options in 2023. In Google Ads, data-driven is the default for most conversion actions and last click is the alternative. Google announced in April 2023 that first click, linear, time decay and position-based would go. From June 2023 they could no longer be selected for conversion actions that were not already using them, and actions still on them switched to data-driven in September 2023.

The GA4 attribution models changed in step. Google Analytics removed the same four rule-based models as of November 2023. GA4 now offers data-driven attribution and two last-click variants: paid and organic last click, and Google paid channels last click. Two GA4 details explain some reporting surprises:

• The default lookback window is 30 days for acquisition key events (first visit, first open) and 90 days for all other key events.

• Data-driven credit can be reattributed for up to 7 days after a conversion, so recent numbers can shift after you export them.

What affiliates are actually paid on

Affiliate attribution works differently from platform attribution. Most affiliate networks and operator programs assign each referred player to one affiliate, typically the one whose tracking link was clicked last before registration, inside the window in the program terms. Some networks add rules on top, such as ranking cookie types or restricting certain publisher types, so read the terms. Awin, for example, lets a click cookie beat a view cookie even when the view came later.

Three things follow from that:

• Programs usually count affiliate links, not your ad platform. Your Instagram ad and the operator's own brand-search ad are normally outside the program's records. In our example, the comparison-page click wins the player.

• Only qualified conversions pay. A registration or first deposit counts once it meets the deal's conditions and passes the program's review. Rejected or reversed conversions never appear as payable, whatever Meta reported.

• The tracker is your record. The program sends approved events back by S2S postback. Your tracker matches them to the original click, and that is the number to reconcile against your payout.

So the conversions Meta or Google report are an optimization signal, not payable conversions. Whatever model an ad platform uses, it does not change what the operator pays.

Last-click payouts also explain why cookie stuffing exists: a fraudster forces their own cookie in front of a real referral so they become the last click. Programs watch for that pattern and can withhold or reverse commissions when they find it.

Why the numbers never match

Your ad platform, tracker, analytics and affiliate dashboard will show different conversion counts for the same campaign. That is normal. Work outward from the platform to find the cause:

• Windows and models. Meta may credit a view or engagement from yesterday. Your tracker credits only tagged clicks, and the program credits its own last affiliate click.

• Repeat counting. A platform can count several conversions from one person, while a CPA deal pays once per player.

• Cross-device and privacy limits. Someone who clicks on a phone and deposits on a laptop, or uses a browser that limits tracking, may never be matched.

• Zero-click influence. Answers shown in search results or AI summaries can shape a decision with no click to track.

• Postback gaps. On the tracker side, a misconfigured or delayed postback means a conversion arrives days late or never.

• Untagged traffic. Clicks without tags land in the wrong bucket, so a paid source can take credit for organic sign-ups, or the reverse.

• Approval lag. On the program side, conversions sit in review and some get rejected, so today's figure changes next week.

• Time zones. Each dashboard closes its day at a different hour.

Decide which number pays the bills. For ROI, use the conversions your program approved and your tracker matched to a click. Use the platform's count for bidding and creative tests.

Incrementality: did the ad actually cause it?

Attribution assigns credit; incrementality asks whether the conversion would have happened without the ad. The two answer different questions, and a campaign can score well on one and badly on the other.

The standard test is a holdout: split the audience, or a set of regions, into a group that sees the ads and a control group that does not, then compare conversions. The difference is the incremental lift. Google offers a version of this as Conversion Lift, which compares people who see your ads with a control group who do not. It is not available to every account. Retargeting and brand search are the usual suspects. They reach people already on their way to converting, so attribution flatters them more than a holdout test does.

Choosing a model in practice

No single model is right for everyone. These rules cover most affiliate setups:

• One channel and a short funnel: last click is fine and sits close to how you are paid.

• Traffic from several sources, with players who take days to deposit: multi-touch or data-driven reports are worth opening, as long as you compare them with what your tracker shows.

• Leave platform attribution settings alone until you can say what each change is meant to fix.

• Write the model and window next to every number you report.

• Reconcile weekly against program-approved conversions, and track value with EPC and related metrics.

FAQ

What are the different types of attribution models?

There are three groups. Single-touch models (first click, last click) give all credit to one touchpoint. Multi-touch models (linear, time decay, position-based) spread it across the path. Data-driven models learn the split from your own conversion data, inside the ad platform.

What is the best attribution model?

It depends on your funnel, and no model wins everywhere. Last click works for short, single-channel funnels and is close to how most affiliate programs pay. Data-driven suits multi-channel accounts with enough conversion volume. Whatever you pick, keep it consistent so reports stay comparable.

What are the four types of attribution?

The four most often listed are first click, last click, linear and time decay or position-based. Lists vary: many add data-driven as a fifth, and some group the rule-based models as single-touch versus multi-touch. Google Ads and GA4 now offer only last click and data-driven.

Can you provide an example of an attribution model?

A bettor clicks an affiliate's Instagram ad, later clicks the operator's brand-search ad, then deposits through the affiliate's comparison page. Last click credits the comparison page in full, first click credits Instagram, and linear gives each touch about a third.

What attribution model do affiliate networks use?

Many affiliate networks and operator programs credit the last affiliate click before sign-up, within the window set in the terms. Some add rules, for example Awin ranks click cookies above view cookies. Affiliate attribution rules on windows and tie-breaks differ, so check each agreement.

Why do my Meta numbers not match my tracker?

They count different things. Meta can credit views and engagements inside its windows and count repeat conversions. Your tracker usually records only tagged clicks and postbacks, which can arrive late or not at all. Reconcile against the conversions your program approved.

Sources

• Google Ads Help, About attribution models

• Google Ads Help, First click, linear, time decay, and position-based attribution models are going away

• Google Analytics Help, Get started with attribution

• Google Analytics Help, Select attribution settings

• Google Ads Help, About Conversion Lift

• Meta Business Help Center, About attribution models and attribution settings

• Meta Business Help Center, About incremental attribution

• Meta for Developers, Ads Insights API metric availability updates

• Awin Help, Awin cookie types and attribution hierarchy