What is multi-touch attribution?

Multi-touch attribution splits the credit for a sale across several touchpoints on the buyer's path. Single-touch models give it all to the first or last click instead. A rule can split it evenly, or a data-driven model can weigh each touch from your own paths. It usually changes little if most buyers buy after one visit.

What one store's data shows

The usual answer sells multi-touch as the fair fix for last click: every touch gets its share. One store's export asks a plainer question first. How many touches, and how many channels, are there to share between?

One store's anonymised GA4 export, 1 January 2024 to 21 August 2026. It holds shares of revenue only: no ad spend, no order counts.

Start with the touches. On the Journeys sheet, journeys with 1 touch hold 79.5% of revenue, with 0.5 days to buy. A multi-touch rule needs two touches or more before it can split anything. On a single touch, linear, data-driven and last click all hand the whole sale to the same place.

Then the channels. The touched view on the Channels sheet gives every channel on a journey full credit, so its total runs to 110.4% by design. Suppose every journey of two or more touches mixed channels. Then the touched total would overshoot by at least the 20.6% those journeys hold on the Journeys sheet. It overshoots by much less.

So a good part of this store's multi-touch revenue came from journeys that kept returning through one channel. That is the catch in the definition: multi-touch is not the same as multi-channel. Five visits through one channel make five touches. Splitting them five ways hands every slice back to that channel, which is fair in the way splitting a pizza with yourself is fair.

The Direct row tells the same story from the other end. On the Channels sheet, Direct holds 57.7% of revenue in the last-click, first-click and touched views alike. Within rounding, the journeys that touched Direct both started and ended on it. A splitting rule could only pass its credit to touches in the middle of those paths.

The longest journeys are where rules disagree most, and they hold the least money. On the Journeys sheet, journeys of 10 or more touches hold 3.0% of revenue, spread over 1,792 distinct path sequences, with 16.0 days to buy. Distinct paths are not purchases, but the shape is plain: plenty of variety, very little revenue.

What the export cannot show is what any touch caused. It holds revenue shares for paths GA4 recorded, so it is silent on spend, orders and the ad someone saw and never clicked. Any split, fair or not, divides sales that already happened.

Why does the textbook version mislead?

The textbook offers a menu: linear, time decay, position-based, U-shaped, W-shaped. Most of that menu has left the tools you actually use. GA4 dropped first click, linear, time decay and position-based in November 2023. Google Ads no longer supports them either, and moved the conversion actions that used them to data-driven.

So in Google's tools, multi-touch now means one model: data-driven. Shopify still keeps a classic rule. Its Linear model gives equal credit to each click in the journey, direct visits included. GA4's models leave direct visits out unless the whole path was direct. One order, two multi-touch answers.

Data-driven is the clever one, and its description sounds like cause. Google says it evaluates both converting and non-converting paths, and contrasts what happened with what could have occurred. It is still a way of dividing credit inside paths Google recorded. Google's Ads help even says that, depending on data availability, last click and data-driven can give the same results.

And every tool splits only what it can see. Google Ads builds its path reports from the keywords and ads in your own Google Ads account. Its help warns that this is why those paths can look shorter than you expect. GA4 sees visits to your site, plus some YouTube engaged views. Neither sees a radio spot, a market stall or a recommendation in a group chat.

For the single-touch side of the story, read what last click attribution is.

What can multi-touch attribution not tell you?

It cannot tell you what a channel added. Every rule, data-driven included, shares out credit for orders that happened. Whether an order needed the ad is a separate question. A holdout answers it: switch the channel off in some regions and compare total sales there.

It cannot rebuild a journey the tracker missed. Someone who heard your name in a group chat and typed it in has a one-touch path, however long their real journey was.

It cannot make your tools agree. GA4, Google Ads and Shopify each split credit their own way, over their own touches. Their totals for one channel will differ, and adding them up counts the same orders twice.

What to do this week

1. Size the revenue a split can reach. In GA4, click Advertising, then Attribution paths under Attribution, set Path length to greater than 1 and click Apply. Pass: multi-touch paths carry a real share of purchase revenue, so the model choice matters. Fail: they carry a sliver, so fix your tagging before you argue about models.
2. Count the paths that mix channels. Sort the same table by purchase revenue and read the top ten paths. Pass: most of them show two or more different channels, so a multi-touch model has something to share. Fail: most repeat one channel, so any split hands the credit straight back.
3. Check that data-driven has enough to learn from. In Google Ads, open Goals > Attribution > Path metrics for your purchase action. It shows how many conversions happened, and after how many ad interactions. Pass: you clear Google's recommended 200 conversions and 2,000 ad interactions in 30 days. Fail: you fall short, so do not be surprised if data-driven looks a lot like last click.

Check the homework. Your GA4 Attribution paths export already holds the evidence. Causality Engine reads that one file and shows what each channel caused next to what last-click gave it, in 1 to 2 minutes, for €99 once (excluding VAT), refundable within 30 days. Check the homework

Sources, 1 October 2026: Get started with attribution (Google); Select attribution settings (Google); Key events attribution paths report (Google); About attribution models (Google); About data-driven attribution (Google); About attribution reports (Google); Marketing reports (Shopify).

Related answers

• All Attribution models questions
• How to read multi-touch attribution in GA4, step by step
• Multi-touch attribution, defined
• Causal attribution, explained