Is multi-touch attribution dead for health and beauty?

Not dead, but half blind for a health and beauty store. If you sell skincare, makeup or supplements, creators often persuade without a click. Multi-touch only splits the site visits GA4 records, so a later search or typed visit takes the credit. It usually still works for those; judge creator and social spend with a holdout test.

If you sell health and beauty products

If you sell skincare, haircare, makeup or supplements, much of the selling happens on someone else's screen. A creator's routine video can answer the questions your product page answers, and leave your site no trace. Viewers turn up later, through a brand search or a typed address, sometimes days after.

GA4's multi-touch model, data-driven attribution, splits only the touches GA4 records. Those are visits to your site, plus engaged views of YouTube ads, such as watching an ad for 30 seconds. A video watched on TikTok or Instagram without a tap is not a touchpoint in GA4 at all.

Then come the reorders. When the serum runs out, regulars come back by bookmark or typed address and buy the same thing. Those journeys are short and single-channel, so every model, multi-touch or not, gives them the same answer.

Here is what one store's export shows. Store A is one store, and its export does not say what it sells.

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.

On the Journeys sheet, journeys with 1 touch hold 79.5% of revenue, with 0.5 days to buy. No model can split a journey with one recorded touch, however clever it is.

On the Channels sheet, Paid Social and Organic Video both hold 0.0% of revenue in every view. The export cannot say why. Either this store ran no social ads and posted no videos, or their viewers arrived without tags and joined other rows.

If you sell beauty products and pay creators, those are the rows to check first in your own export. A channel at zero in the paths gets zero from every model. Multi-touch attribution cannot share credit with a channel it never recorded.

What changes for a health and beauty store?

• Views do the work, clicks collect the credit. Meta counts some views itself. Its engage-through setting includes video plays of 5 seconds, and view-through counts events within 1 day of an impression. GA4 sees none of that without a click, so the two disagree most on video.
• Creator links need one naming rule. Google asks for one utm_source per platform, such as Instagram or TikTok. Put the creator's handle in utm_content, and GA4 reports it as Session manual ad content.
• Consent gaps hurt smaller brands more. When a visitor declines analytics cookies, GA4 cannot join their visits into a path. Google models the gap only for properties with at least 1,000 daily consenting users on 7 of the previous 28 days.
• Reorders need no model at all. Replenishment journeys tend to be short, so multi-touch matters mostly for first orders, where buyers compare shades, formulas and prices.
• Meta's incremental column is a second opinion. It counts only the conversions Meta considers caused by its ads, so it sits at or below your standard results.

What to do this week

1. Find your creators in GA4. In Explore, start a Free form exploration with Session manual ad content as the dimension and Sessions and Purchase revenue as metrics. Pass: every paid creator's handle has a row. Fail: some are missing, so their links go out untagged and their buyers land in other rows.
2. Ask Meta which video ads it doubts. In Ads Manager, open Columns: Performance, choose Compare attribution models, select Incremental and click Apply. Pass: your creator and video ads keep most of their results under Incremental. Fail: they lose most, so Meta itself is unsure they caused those sales; test before you scale them.
3. Hold creator ads out of some regions. If you boost creator posts as paid ads, set Locations at the ad set level to half your regions. Untick Reach more people likely to respond to your ads. Pass: you sell in enough regions to split them into two similar groups. Fail: you sell in one small market, so run the ads in alternate fortnights and compare first-time customers in Shopify's New vs returning customers report.

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); Key event attribution paths report (Google); About attribution models and attribution settings (Meta); Collect campaign data with custom URLs (Google); Traffic-source dimensions, manual tagging, and auto-tagging (Google); Consent mode on websites and mobile apps (Google); Behavioral modeling for consent mode (Google); How to view results for incremental attribution in Meta Ads Manager (Meta); Free-form exploration (Google); Use location targeting (Meta); Customers reports (Shopify).

Related answers

• All Attribution models questions
• Is multi-touch attribution dead?
• How to check if multi-touch attribution still works for you
• Incrementality testing, explained