Causal attribution for ecommerce brands

Most attribution does not ask that question. Platform dashboards and last-click models hand out credit by rules the platforms wrote themselves, and the report is reliably flattering to whoever wrote it.

Causality Engine is causal attribution software built for ecommerce and DTC brands. We read every channel GA4 records, paid and organic alike: search, social, email, referral, and the Direct traffic we split back to the channels behind it. Not just last-click ROAS. Upload a GA4 export, from Shopify or any other store platform, and we return what each channel caused, next to what last-click gave it, with a data-health score on every channel, in 1–2 minutes. No pixel, no SDK, no integration project.

Incremental marketing, holistically. Not just paid ROAS.

Paid-channel ROAS is the first place teams look, but it is not the whole picture. Organic search created demand. Email captured it. Brand recognition shortened the consideration window. Retention spend defended the lifetime value. Each activity has an incremental component, and most of them are invisible to a last-click dashboard.

Causal attribution treats marketing as one connected system and estimates what each channel the data records contributed: paid, organic, email, referral. The question is always the same: what would have happened without this activity? The answer is always a number you can check against your own export, never a guess in a suit.

Platform attribution tells you who to thank. Causal attribution tells you what to cut.

Last-click gives the sale to whatever the buyer touched last. Multi-touch spreads credit by a formula you did not choose. Each ad platform counts the same conversion in its own dashboard, so your channels can add up to more than the orders you actually shipped. They cannot all be right.

That overcounting stays invisible until you measure incrementality. The practical cost is that budget moves to the channel reporting the biggest number, not the channel causing the most sales.

How causal attribution works

Causal attribution treats your marketing like an experiment that already happened. Instead of trusting each platform's self-report, it estimates the counterfactual (what sales would have been without a given channel), using the converting paths already in your GA4 data.

The output is what each channel caused, next to what last-click gave it: credit earned, not credit claimed. Because the method leans on causal inference (the same statistical machinery used to evaluate medicine and policy), every channel ships with a data-health score, so you know how far to trust the read before you move money.

What you get

• What each channel caused, not just paid channels

  Each channel's share of the credit once Direct is split back to the channels that sent those buyers: paid, organic, email and the rest.

• Wasted spend and under-invested channels surfaced

  Both the budget propping up channels that are not causing incremental sales and the under-invested channels that are quietly driving demand.

• A data-health score on every channel

  A score from 0 to 100 that grades the read, not the channel. Under 90, move budget in steps.

• A plain-English report

  What to scale, what to cut, what to test next. No statistics degree required.

From data to decision in three steps

1. 01

   Upload

   Export 12 months or more of GA4 data, from Shopify or any other store platform. Longer reads better. No pixel to install, no tag manager, no engineering ticket.

2. 02

   Analyze

   The causal model runs in 1–2 minutes and estimates what each channel caused.

3. 03

   Decide

   You get what each channel caused, next to last-click, and a next step per channel: where to raise, cut or hold.

Why brands trust the numbers

We document every assumption: prior, functional-form choice, covariate, robustness check. A causal claim is only as good as the assumptions under it. Our methodology document is available to any customer who asks. The goal is not a prettier dashboard. It is a number you can defend in a budget meeting.

Causal inference is the machinery economists use to evaluate policy, and it has its own vocabulary: potential outcomes, d-separation on a causal graph, and the checks for spurious correlation, regression to the mean and exogeneity. None of those words appear in last-click attribution, because last-click does not measure causation. For the long version, Hernán and Robins' Causal Inference: What If (Chapman & Hall/CRC, 2020) is free to download.

Who it is for

Ecommerce and DTC brands on GA4, whatever the store platform: the owner who signs off the ad budget and the marketing manager who has to defend it. If you are making budget decisions off three dashboards showing three different numbers, causal attribution is the tie-breaker.

Where this fits in the causal stack

• Incrementality testingThe same incrementality question, answered from your existing data, with no geo holdout.
• Marketing mix modelingWhat a mix model tells you, and the faster read from a GA4 export.
• Attribution in plain EnglishThe short version, without the jargon, for whoever signs off the budget.
• Shopify attributionCausal attribution for Shopify stores, from the GA4 export.
• GA4 attributionCausal, not last-click, on your GA4 export.
• Pricing€99 per causal read. No subscription.
• Average treatment effectThe statistical object causal attribution estimates.
• Causal forestsA modern machine-learning approach to heterogeneous treatment effects.
• Causal chainHow influence flows from awareness to conversion.
• Collider biasThe trap that breaks naive multi-touch attribution.
• Spurious correlationWhat every overcounted dashboard mistakes for causation.