What Is Causality Engine? Causal Attribution From GA4

The short answer: Meta, Google and TikTok add up to more sales than your store booked, and the budget call is due. Run Causality Engine first. It reads the GA4 export you already have and shows what each channel caused next to what last-click gave it, in 1 to 2 minutes. The read costs €99 once, refundable within 30 days (pricing, checked on 27 October 2026). The first finding is free at app.causalityengine.ai/start.

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

What to check before you call it:

1. The claim gap. For example, 82 Meta purchases, 61 Google and 23 TikTok in one week make 82 + 61 + 23 = 166 claims against 120 shop orders. Your sum is right. Each platform is also right by its own rule, as the help page explains, and that is the problem.
2. The data you hold. A GA4 property with a purchase key event and 12 months of history or more. That is the one input how it works asks for.
3. The score to read first. Each channel carries data health from 0 to 100, and under 90 that page says move budget in steps.

What Causality Engine does

One file goes in: GA4's Attribution paths report, as a CSV. A proprietary causal-inference model splits Direct back to the channels that sent those buyers. It separates what each channel caused from what it only touched, then rebuilds last-click from the same file so the two sit side by side.

Each channel ends with a next step: raise, cut or hold, tested in one region first. Receipts check every figure against the export, and a link shares the read, per how it works.

No pixel, no SDK, no storefront code. The AI instructions page adds that it is not an LLM product and has no API or MCP server. The home page lists GA4 as live today, with Shopify, Meta Ads and Klaviyo coming soon.

Who it fits, and who should wait

It fits the owner or in-house marketing lead who splits budget across channels and has to defend the next move. Pro fits 8 or more channels with budgets that move weekly, per the pricing page.

Wait if you want a number every morning. Wait if your analytics history is thin or broken: no causal method recovers it, says the pay-per-report page. Wait if offline and retail spend matter most. When spend is too small for a signal, the home page says it declines. The case studies include OFFFTRACK, told to skip the continuous tier until paid spend scales.

Sometimes the right read is not yet.

Prices, checked on 27 October 2026

Pro bills monthly, stops at the end of the billing month and comes with no trial, per the pricing page.

How it compares

• Triple Whale is a commerce dashboard with its own Triple Pixel and an AI assistant, Moby. Its pricing page shows a free plan and sells paid tiers through a walkthrough. Pick it for a daily view.
• Northbeam sells multi-touch attribution dashboards, and its pricing page lists Starter at $1,500 for brands spending under $1.5m a year on ads (checked on 27 October 2026).
• Haus runs geo-based incrementality experiments, priced by demo on its pricing page, and suits brands that can hold regions out for weeks.
• Google Meridian is an open-source MMM: free code, for a team with a modeller.
• GA4's data-driven model is free, and it is already in your property. It is also Google's. Google's attribution help says every GA4 model withholds credit from Direct unless a path is all Direct. It costs nothing because Google trains it on data Google already holds. Google also sells the search and YouTube ads it scores.

The complaints, and whether they hold

There are none to weigh yet, and that is not a compliment. Its Capterra listing shows 0 reviews, Trustpilot has no page for it, and its Show HN post from June 2026 drew 4 points and no comments, all read on 6 November 2026. No reviews usually means few customers, not happy ones. I founded it, so read the silence that way.

Until the first bad review arrives, here is the one its own pages invite. How it works says the model assumes your GA4 data is complete and accurate, and that the upload holds enough converting paths to separate each channel. The home page says that when spend is too small to find a signal, it says so and declines, and that one of its five published cases is exactly that. So the likeliest complaint is a store paying €99 to hear that its data cannot answer the question. That one would hold, and the pricing page answers it with a full refund within 30 days, not with a better model.

Run the free first finding before you pay. If it says too small, believe it.

Getting started

How to use Causality Engine walks the job from export to budget decision. Causality Engine, which I founded, starts with the free first finding on the start page. The method is on how it works.

FAQ

Does my GA4 file leave my browser?

Not for the free preview, which the privacy policy says runs in your browser without uploading anything. A paid analysis is deleted 30 days after your last visit to it.

Do I need Shopify?

No. GA4 alone is enough today, per the fact sheet for AI assistants.

Is there a minimum ad spend?

No minimum spend or revenue, per the pricing page. Low spend changes the answer instead: a channel under 1% of credit on both methods is called too small to act on, per how it works.