From CSV to causal answer in 1–2 minutes.

1. 01

   Upload

   Export 12 months or more of GA4 data. Longer reads better. The read takes one file, the Attribution paths export from GA4 as a CSV, whatever your store platform. Drop it into Causality Engine. On Pro, GA4 is ingested automatically.

2. 02

   Analyze

   Our causal-inference model reads the converting paths in your export, splits Direct back to the channels that sent those buyers, and separates what each channel caused from what it only touched. It rebuilds last-click from the same file, so you see both. A read finishes in 1–2 minutes.

3. 03

   Decide

   You see every channel next to last-click, with a data-health score from 0 to 100 and a next step: raise, cut or hold, and how to test it in one region first. Receipts check every number against your export, and a link shares the read with whoever signs off the budget.

What the model assumes (and what it does not)

Every causal claim rests on assumptions. Ours are documented, not hidden. The model assumes your GA4 data is complete and accurate, that the window you upload holds enough converting paths to separate each channel, and that no factor your data cannot see moves sales in lockstep with one channel. Data health grades how far each channel's read holds up: under 90, move budget in steps. A channel under 1% of credit on both methods is called too small to act on, not given a verdict.

It does not assume your platforms are honest. It does not assume your last-click rule is right. It does not assume causation from correlation. And it shows its work: every number in the read checks against your export.

Why no pixel

Pixels have been steadily degraded by iOS privacy changes, third-party cookie deprecation and platform-side modeling. Patching them with modeled conversions inside each platform reintroduces the platform's bias. We work from the GA4 data you already collect instead, under the same consent and browser limits as GA4, without adding another tag to your store.