Causality Engine vs Windsor.ai: Moving the Data or Reading It

The short answer: getting every platform's numbers into one place is one job. Deciding which of them caused the sales is another. If the job is getting Meta, Google Ads and GA4 numbers into one sheet, dashboard or AI chat every morning, Windsor.ai does it. It costs from $19 a month billed annually, or $23 monthly, for 3 data sources (Windsor.ai pricing). If the job is deciding which of those channels caused your sales, Causality Engine is the pick. One read of the GA4 export costs €99, and the first finding is free on the start page (pricing page). Both checked on 27 October 2026.

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

A connector is very good at one thing: carrying each platform's numbers somewhere you can see them together. That is usually when the trouble starts. The columns often add up to more revenue than the store took. Windsor.ai moves the data. Causality Engine reads it for cause.

Size the job first:

1. What a year of plumbing costs. Windsor.ai Standard is $118 a month on monthly billing, for example: 12 x $118 = $1,416 a year. Billed annually it is 12 x $99 = $1,188, a saving of $228, per the pricing page. One causal read is €99, once, on Causality Engine's pricing page.
2. What you already move. If Windsor.ai already lands spend by channel in a Google Sheet, keep it. The read's caused share per channel becomes one more column beside that spend.
3. What the free month covers. Windsor.ai's free plan starts with a 30-day trial of 10 data sources and 15 accounts. Then it steps down to 1 source and 1 account with 30 days of history (pricing page). Note which connections earned their place before day 30.
4. What would change your mind. Pick the channel whose share of spend and share of caused revenue differ most. Test it in one region before moving the whole budget, as Causality Engine's how-it-works page advises for every channel.

What each one hands you

The differences that matter

A faithful copy carries the overcount with it

Windsor.ai's answers page is clear about its numbers. When they differ from Ads Manager, Google Ads or GA4, the cause is usually a different query. So match the level, the date range, the time zone and the attribution setting. That is the right promise for a connector. It also means each platform's attribution arrives intact, including the sale that Meta and Google both claim. Summing the columns in a sheet is the obvious next step. It reproduces the overcount faster, and resolves nothing.

Two meanings of attribution

Windsor.ai's own post on data-driven attribution recommends moving from last click to multi-touch or data-driven credit, and judging channels on a data-driven ROAS. Shared credit across a journey is a real improvement on last click. Causality Engine asks a different question of the same journeys: what sales would have been without each channel. It splits Direct back to the channels that sent those buyers and publishes the assumptions under each estimate (causal attribution).

What a decision needs that a dashboard does not

A refreshed dashboard is the right tool when the team reads numbers every morning. Causality Engine's own pay-per-report page says so: if numbers are read daily and budgets move weekly, a dashboard earns its price. A budget split is different. It is a decision with an end, and it needs a number per channel with a confidence attached, not a fresher copy of each seller's claim.

An agent in the loop

This job suits an agent, with a person in front of it. Connect Windsor.ai to Claude through Windsor MCP and ask for spend by channel for the same months as your GA4 export. Connect each ad account with a platform user that has read-only or reporting access, so the platform rejects any write request, as the answers page recommends. The human check: before the agent's totals go next to the read, compare one month against each ad account's own spend screen. Then the sheet holds spend from Windsor.ai and caused share from Causality Engine. Nothing in it was changed by the agent.

Plans and terms, 27 October 2026

Windsor.ai, in US dollars on its pricing page: Basic $19 a month billed annually ($23 monthly) for 3 sources and 75 accounts; Standard $99 ($118); Plus $249 ($299); Professional $499 ($598); Enterprise quoted. Every plan includes all sources and destinations, and Windsor MCP.

Causality Engine, excluding VAT, on its pricing page: €99 per read, refundable within 30 days. Pro, €299 a month and billed monthly, brings automated GA4 ingestion, budget alerts and unlimited seats. Enterprise is a quoted 90-day pilot.

Windsor.ai's answers page sets the terms of leaving. Subscriptions renew automatically, all sales are final, and cancelling deletes the historical data it holds. So the archive you build inside the tool is also what keeps the renewal going. Point every sync at a Sheet or warehouse you own, from the first month.

Causality Engine is the call when

• The question is which channel caused sales, and the spend data already sits in a sheet you trust.
• You want an answer you can defend to finance, with each channel's health score showing how far the number carries.
• You need it this week from the history GA4 already holds, with no connector to configure.

Windsor.ai is the call when

• Data from many platforms and client accounts has to reach Looker Studio, BigQuery or an AI chat on a schedule.
• An agency reports on dozens of accounts and needs destinations included rather than priced one by one.
• You want an AI assistant that can pause campaigns or change budgets, with confirmation on every change. Causality Engine moves no budget anywhere.

Background: the Causality Engine glossary entry, the Windsor.ai glossary entry, and how to use Windsor.ai for the connector side.

One free check, one paid answer

The free check needs no new tool. Put last month's platform-reported revenue per channel, from the Windsor.ai sheet you already have, next to the store's real revenue. If the columns sum to more than the store took, the overcount is on the page. Then drop the GA4 Attribution paths export on the start page. The file stays in your browser while the preview reads it, per the privacy policy.

The paid answer is the full read in Causality Engine, which I founded, from the start page: €99, Direct split back, and a raise, cut or hold per channel with a one-region test, as how it works describes. If your spend is too small for a signal, the homepage says the read declines rather than invents one.

FAQ

Can Windsor.ai feed Causality Engine directly?

Neither side documents a connection. The read takes the Attribution paths CSV you export from GA4 (how it works). Causality Engine's homepage lists GA4 as its one integration live today, with the rest marked coming soon.

Does Windsor.ai do attribution?

It offers an Ad Data + Attribution connector and dashboard templates for Looker Studio (template page). Its own writing favours data-driven, multi-touch credit. Causality Engine's read estimates what each channel caused instead.

What happens to my Windsor.ai data if I cancel?

Its answers page says cancelling deletes your historical data, and that there is no pause option. Keep a copy in a destination you own.