Causality Engine vs Haus: Which Geo Test to Run First

The short answer: sell from the Netherlands, or from anywhere outside the US, and Haus's self-serve Shopify plan is closed to you. Its pricing page lists Basics as US only. International testing starts at Core. For that brand, and for any brand that can afford only a few multi-week tests a year, the first move is a Causality Engine read of the GA4 export. The first finding is free and the full read is €99 once (pricing, checked on 27 October 2026). It tells you which channel deserves a geo test. Choose Haus when you can hold ads back in some regions for weeks at a time. It carries those experiments into a weekly mix model, with a strategist beside you.

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

Three rules for the test calendar:

1. Count the weeks before you count the money. Haus's guide to test duration gives working baselines. Branded search needs 2 or 3 weeks, Meta conversion campaigns 3 to 4, Pinterest and Snap 4 to 5, Meta reach 6 to 8. Run Meta reach tests back to back at 8 weeks each, for example, and they fit 52 ÷ 8 = 6.5 times into a year. So six. And only if nothing else needs those regions.
2. Start from the file you have. The read takes one export, the GA4 Attribution paths report as a CSV with twelve months or more in it. It touches nothing on the store, per the how-it-works page.
3. Write down the review date before the test starts. Each channel in a read ends on a next step with a one-region test, and the sample read on the pricing page checks back after 14 or 21 days. Fix the revenue line and the date first. Then touch the budget.

What Haus sells

Haus is an incrementality platform with three products, per its home page: experiments, a Causal MMM updated weekly, and Causal Attribution for daily calls down to the ad. Its experiments page lists three kinds of test. GeoLift covers any channel whose spend can vary by geography. Fixed Geo Tests run in regions you choose. Time Tests cover one-off moments such as a product drop. PhD scientists interpret the results. On top sits Architect, which Haus calls its causal marketing agent: it runs what-if analyses and recommends the next move. Treat its budget shifts as proposals, approved by the budget owner before they go live.

Causality Engine sells one thing at two speeds: a causal read of your GA4 export, once or continuously on Pro.

Side by side

Two things that decide it

A test answers one question, a read ranks the questions

A GeoLift test is the real thing. Haus assigns test and control regions by random stratified sampling and measures the difference, and for the channel under test nothing observational comes close. It is also slow. At six Meta reach tests a year, the order of the tests matters more than any one result. A read sets every channel side by side: what it caused next to its last-click credit, each with a data-health score from 0 to 100. The widest gap on the biggest budget goes to a geo test.

Geography is the entry ticket

Geo testing needs regions where spend can be switched up or down and sales compared. Haus's Basics plan does this in the US. Outside it, international testing begins on Core, priced through a demo. The read has no regional requirement at all. It reads paths, not regions.

Picking the first test

The order that wastes the fewest weeks is short. Take the read. List the channels where engine and last-click disagree by more than a quarter (the sample on the pricing page holds anything within 25%), then sort them by spend. The top one goes to a geo test: on Haus, on Meta's open-source GeoLift (a research tool, by its own description), or by hand. A step-by-step version is in how to run a geo holdout test.

Causality Engine's incrementality page adds a reading rule for whatever test you run. An interval well above 1.0x means the channel pays for itself. One that straddles 1.0x means it may not be incremental at all.

What each costs

Checked on 27 October 2026. Causality Engine's pricing page: €99 for a single read, paid once and refundable within 30 days. Pro is €299 a month, billed monthly, with automated GA4 ingestion and a chatbot over your own attribution data. Enterprise is a custom 90-day pilot. VAT is excluded throughout. Haus's pricing page shows Basics for Shopify, Core, Plus and Enterprise, each behind a "Get a demo" button, with optional add-ons such as international support and extra Causal MMM models.

Haus's free Shopify app starts by installing the Haus pixel, which collects page views, carts and purchases, per its App Store listing. So the door marked free puts Haus's tracking on your store before any plan has a price. Causality Engine adds no tag. Its how-it-works page says it reads the GA4 data you already collect, under the same consent limits as GA4.

The free route

Start free. Run the first finding on the start page; Causality Engine's home page says the file never leaves your browser. Sketch a year of tests on one sheet with Haus's duration baselines. If one channel stands out, run a single holdout yourself, on GeoLift or by switching spend off in a few comparable regions, and judge it by the interval rule. Pay the €99 when the free finding shows Direct hiding enough to matter. Take Pro when eight or more channels and weekly budget changes make a quarterly read too slow.

Choose Causality Engine if

• Your job is deciding where next month's budget goes, from spend you have already made.
• You sell outside the US and want an answer before you commit to a plan priced by demo.
• You can fund only a few experiments a year and need to know which one to run.
• You do not want another pixel on the store.

Choose Haus if

• Your job is an experiment programme, with a strategist choosing tests and reading results.
• You can hold spend back in regions for 2 to 8 weeks per test without hurting the quarter.
• You want results fed into a weekly Causal MMM and daily attribution.
• You measure influencer, podcast, out-of-home or TV, which Haus covers on Plus and Enterprise.

Both can run at once. The read ranks, the experiment confirms. Causality Engine's own comparison with Haus, the company's view, says to keep Haus and add the read.

Rank the channels before the regions

Causality Engine, which I founded, gives you the ranking first: drop the export on the start page for the free finding, then pay €99 for the full read, with the method set out on the how-it-works page. The glossary has a plain description of Causality Engine.

FAQ

How long does a Haus test run?

It depends on the channel. Haus's duration guide suggests 2 or 3 weeks for branded search, 3 to 4 for Meta conversion campaigns, and 6 to 8 for Meta reach. The usual minimum is 14 days.

Can a brand outside the US use Haus?

Yes, from Core up. Basics for Shopify is US only, per Haus's pricing page.

Does Causality Engine run geo experiments?

No. It reads past GA4 paths and suggests, for each channel, a one-region test to confirm the next step, per its how-it-works page. Running that test is your job, or Haus's.