Become an affiliateStart your program

Technology and tracking

Multi-touch attribution

Also called multi-touch attribution, MTA, attribution models.

Short answer

Multi-touch attribution is an attribution method that splits the credit for a conversion between several touchpoints in the customer journey – for example 40 % to the first click, 40 % to the last and 20 % to those in between. The method is used for analysis; paying out commission requires an unambiguous model.

Definition of Multi-touch attribution.

The classic models split credit by position or time: first-click gives everything to the first touchpoint, linear splits it evenly, time-decay weights the most recent touches highest, and position-based (U-shaped) typically gives 40 % to the first and last click and spreads the rest in between. Data-driven models let an algorithm weight touchpoints from historical conversion data. What they all have in common: the result depends on the choice of model, not just on the data.

Multi-touch assumes that the whole customer journey can be observed – every visit must be linked to the same person across channels, browsers and devices. That requires login data, fingerprinting or identity graphs. With ITP, cookie consent and several devices per person, the complete journey is in practice a fragment, and the model splits credit based on the fragment it can see.

For analysis, multi-touch can be useful: it shows which channels open customer journeys and which close them. For settling commission it is hard to defend. A payout requires an unambiguous, verifiable answer to who triggered the sale – and a model weight of 23 % is a choice, not a fact an affiliate can check.

Multi-touch vs. last-click attribution

AspectMulti-touchLast click
Creditsplit across several touchpoints according to a weighting modelthe whole conversion is credited to one click
Data requirementthe whole customer journey across devices and channelsa click ID that follows the conversion
Verifiabilitythe weights are a model choice and cannot be checked from outsidea click row documents the attribution
Typical usechannel analysis in the advertiser's own reportssettling commission in affiliate programs

Use multi-touch to understand the customer journey – use last click when money is to be paid out on the result.

How it goes wrong

  • 01

    The model moves the truth

    Switch from time-decay to position-based and a channel's credit can change significantly without a single sale having moved. Conclusions about what works must always be read together with the choice of model.

  • 02

    Payouts by weights cannot be verified

    If an affiliate gets 23 % of a commission because a model weighted the touch that way, there is nothing the affiliate can check. Disagreements cannot be settled, because the answer depends on the model – which is why affiliate commission is in practice settled by a deterministic rule.

  • 03

    The data has holes

    Multi-touch assumes full visibility of the customer journey, but ITP, declined consent and device switching cut journeys apart. The model then splits credit from fragments – and presents the result with the same confidence.

At Affilyflow

Multi-touch attribution in practice

Affilyflow uses one attribution model: the conversion is credited to the click ID it carries. The attribution window is locked at the click and decided on our server – not in the browser – and the same order can never count twice, because orders are deduplicated on order or invoice ID per program. Multi-touch and cross-device tracking have been deliberately left out: a commission must be documentable with a click row that both store and affiliate can have explained.

Frequently asked questions

What is multi-touch attribution?

Multi-touch attribution splits the credit for a conversion between several touchpoints in the customer journey instead of giving it all to one click. Well-known models are first-click, linear, time-decay and position-based. The method is mainly used for channel analysis by advertisers.

Why do affiliate programs use last click instead of multi-touch?

Because money is paid out per conversion, the answer to who triggered the sale must be unambiguous and verifiable. Last click – in practice the click ID the conversion carries – gives a documentable answer. Multi-touch weights are model choices an affiliate cannot check, which makes settlement and disputes unmanageable.

What attribution models are there?

The most common are last click, first-click, linear, time-decay, position-based (U-shaped) and data-driven models, in which an algorithm weights the touchpoints. Affiliate settlement uses a deterministic model, while the others are mainly used in analytics and advertising tools.

Can multi-touch and affiliate marketing be combined?

Yes, but on separate shelves: the advertiser can analyse the affiliate channel's role in the customer journey with multi-touch in its own reports, while the commission itself is settled by an unambiguous rule. Analysis and settlement have different requirements – insight may depend on the model, payouts must be deterministic.

Read on