How to Measure TikTok Ads Incrementality With GA4 Paths

The short answer: TikTok Ads Manager shows a healthy ROAS. GA4 shows TikTok closing almost nothing. Both screens can be right, because they count different events. TikTok's attribution windows run from 1 to 28 days after a click and offer view-through windows of off, 1 or 7 days. So a purchase by someone who only watched can count as TikTok's. GA4 records TikTok only when someone taps through. For a Shopify brand spending on TikTok beside Meta and Google, my recommendation is Causality Engine. Its read of a GA4 export of twelve months or more returns TikTok's caused share beside its last-click share in 1 to 2 minutes, per the how-it-works page. The first finding is free on the start page. The steps: split TikTok's columns, tag the ads and find TikTok in the GA4 paths. Then check TikTok's own reconciliation page, test by region and run the read.

Disclosure: I founded Causality Engine and own it, and this page recommends it. Every claim about it links to Causality Engine's own pages.

Put three numbers side by side first:

1. TikTok's own split. TikTok's attribution metrics glossary separates CTA Purchase, credited to a click, from VTA Purchase, credited to an impression. For example, Ads Manager claims 140 purchases and 96 of them sit under VTA. Then 96 / 140 = 68.57% of the claim rests on views GA4 never sees.
2. GA4's count. Purchases with a TikTok touch anywhere in the path, from the paths report in Step 3.
3. The till. Store revenue for the same dates. It is the one figure neither platform sets.

Step 1: Split the conversions column

In Ads Manager, add custom columns for CTA Purchase and VTA Purchase beside your usual conversions. If you run the Sales objective, add engaged view-through too. TikTok's EVTA article credits an ad when someone watched at least 6 seconds, or all of a shorter video, before converting. It folds those into the Conversions column under your settings, and says custom columns break the three apart. The VTA metrics are counted on impression time, per the glossary, so their dates will not line up with GA4's.

Step 2: Tag every ad so GA4 can name it

TikTok's UTM article lists the five parameters, plus macros such as __CAMPAIGN_NAME__, __AID_NAME__ and __CID_NAME__ that fill in campaign, ad group and creative names. It also recommends turning on the option that adds UTMs automatically. GA4's default channel group then files the visits under one of its paid channels by source and medium. Check which one in Traffic acquisition, then keep the tags identical across campaigns, so TikTok stays one row in every report that follows.

Step 3: Find TikTok in the GA4 paths

Open Advertising, then Key event attribution paths, and switch the table to Source. Google's paths report guide shows paths up to 20 touchpoints and splits each into early, mid and late segments, and lets you filter by path length. Set the filter to more than one touchpoint and look at where TikTok sits. A channel that keeps turning up early, in paths that end on Paid Search or Direct, is doing discovery work that last-click will never pay for. It plays midfield, and last-click only counts whoever touched the ball last. The Days to key event column tells you how long buyers take. GA4's attribution settings keep purchases in view for 90 days by default, with 30 or 60 as the alternatives.

Step 4: Open TikTok's own reconciliation page

TikTok built a page for exactly this argument. Under Analytics, then Attribution Analytics, its reporting comparison article describes a breakdown of TikTok web purchases across several windows, set next to Google Analytics values. It adds a check of utm_source, utm_id, utm_medium and utm_campaign. It needs Google Analytics connected in Events Manager, covers web purchases only, and matches campaigns through utm_id. Use it to fix tagging gaps. The page is useful, and it explains the difference between the two counts. It does not tell you which sales TikTok caused.

Step 5: Test it by region, by hand

TikTok's location targeting article is blunt: TikTok does not support excluding locations. You target the regions you want instead, so build the test that way. Pick regions holding about a quarter of your orders as the off group, and set every TikTok ad group to target only the remaining regions. Then compare store revenue per region in Shopify's billing location report, each group against its own weeks before. The same article warns that very small areas cut delivery and can raise cost per action, so use whole provinces or states, not postcodes.

Run it for four weeks or more, and watch Meta and Google revenue in the off regions too. Causality Engine's TikTok page argues that TikTok creates demand other channels collect. By the same page, cutting it on last-click can pull Google and Meta down a few weeks later. A two-week test can miss that second effect.

TikTok's stricter version is the Conversion Lift Study. It splits people into a group that sees your ads and one that does not, and reports incremental conversions and iROAS. It is also a managed service, for accounts that meet a minimum spend.

Step 6: Run the read on the export

Download the paths report as a CSV and drop it on the start page. The free finding is the number Causality Engine's checkup page describes: how many days buyers who saw two or more channels take to convert. The full read gives TikTok's share of credit after Direct is split back, beside last-click, with a data-health score and a next step. No TikTok connection is involved. The home page lists GA4 as the integration live today and TikTok Ads as coming soon.

TikTok's view-through count runs generous. GA4's tap count runs stingy. Neither is wrong by its own rules, and neither is the number to budget on. Causality Engine's TikTok-first page works an illustration in its own round numbers, not a customer result. €5,000 of spend shows €4,000 of last-click sales, 0.8x. Switch TikTok off and total sales fall by €15,000 a month, so it caused 3.0x. Your read may land above or below last-click. That is the point of running it.

Where the documentation tells you to be careful

• Dates. VTA counts on impression time, GA4 on the purchase.
• Coverage. TikTok's comparison page includes only campaigns it can match through utm_id.
• Weak reads. Under a data-health score of 90, the how-it-works page says to move budget in steps. Under 1% of credit on both methods, a channel is too small to act on.

What the default view setting buys TikTok

Switch a view-through window on, and a purchase by someone who scrolled past the ad without tapping can be reported as TikTok's sale. The Conversion Lift Study that would test that claim is designed and run by TikTok's own team, for accounts above its spend minimum. Trusting the Ads Manager figure was not naive, since it counts views the tap count cannot see. It is still a platform marking its own homework, by design. The regional test above is the one TikTok does not run for you.

The free work first, then the read

Steps 1 to 5 cost nothing but time and the spend you hold back. Causality Engine, which I founded, reads TikTok against every other channel from the same GA4 export. The first finding is free on the start page. The full read is €99 once and refundable within 30 days. Pro is €299 a month with budget alerts when a channel's performance shifts, per the pricing page (checked on 27 October 2026, excluding VAT). The Causality Engine entry has the product in brief.

FAQ

Should I switch view-through attribution off?

It is set per ad group. TikTok's attribution windows article presents it as part of the ad group's strategy, not only a reporting view. Off brings Ads Manager closer to GA4. It does not measure what TikTok caused either.

Does this cover TikTok Shop?

Causality Engine's TikTok page says yes, as long as TikTok Shop purchases flow into your GA4 data.

How much history should the export hold?

Twelve months or more, per the same page, because beauty and fashion buyers can research for weeks after first seeing a product on TikTok.