¶ A/B tests - store experiments with Google Analytics 4

A/B tests in Hummerce let you check which version of a store page or element (e.g. button text) better serves a business goal - higher conversion, more engagement or more sales. The built-in experiments module runs on the store's own resources, with no extra paid tools, and you analyze the results in Google Analytics 4.

Last updated: 2026-09-28.

¶ Benefits and use cases

Validate a change before rolling it out for good - e.g. whether a different call-to-action button text ("Add to cart" -> "Buy now") raises conversion.
No extra paid tools - the experiment runs on the built-in Hummerce module, and the result goes to the Google Analytics 4 you already use.
Random, consistent variant assignment - a customer assigned to a variant once sees it consistently on later visits.
Analysis in a broader context - GA4 data can be moved into BigQuery and combined with other sources (Google Ads, Facebook, the store, CRM, ERP).

¶ How it works

¶ Step 1: define the experiment

Before starting a test, you decide exactly what will be measured (e.g. clicks, add-to-cart actions, purchases), and prepare the variants to compare (e.g. two versions of a button or a page).

¶ Step 2: assign a variant to the customer

On the first session, the store checks whether the customer already has an experiment variant assigned. If so - they see the same variant as before. If not - the system randomly assigns them one of the variants.

¶ Step 3: send data to the DataLayer and GA4

Information about the assigned variant goes into the page's DataLayer, and from there to Google Analytics 4 - this lets you link the test variant to the customer's behavior.

¶ Step 4: analyze the results

GA4 shows the test results in real time and lets you compare variants. GA4 data can also flow into BigQuery, enabling analysis alongside other data sources (Google Ads, Facebook, the store, CRM, ERP). This is the basis for deciding whether to roll out the chosen variant for good.

¶ Dependencies

Sending results requires a configured Google Analytics 4 (Google Analytics 4).
Combining results with other data sources requires exporting GA4 to BigQuery, e.g. through Looker Studio (Reports - BigQuery and Looker Studio).

¶ Implementation

Every A/B test requires its own implementation - preparing or extending the random variant assignment mechanism, verifying that data reaches the DataLayer and GA4, and finally analyzing the results.
Minimum estimated time:
- assigning an experiment variant to the user - from 5 h,
- sending information to the DataLayer - from 4 h,
- sending data to GA4 - from 4 h,
- analyzing the results - from 10 h.
What the time depends on:
- whether the experiments module needs extending, or just configuring an existing mechanism,
- the number of data sources included in the analysis (e.g. integration with BigQuery, Google Ads, Facebook, CRM, ERP),
- the expected depth of the analysis (e.g. segmented comparisons, forecasts).

¶ Limitations

We do not verify the configuration for legal compliance - the store decides the test content and its compliance with the store's terms of service.

¶ Frequently asked questions

Do A/B tests require additional, paid tools?
No. Hummerce has a built-in experiments module that uses the Google Analytics 4 the store already has.

Does the customer see the same test version the whole time?
Yes. Once a variant is assigned on the first session, the customer sees it consistently for as long as it stays assigned to them.

Can test results be combined with other data, e.g. from Google Ads or a CRM?
Yes, if the GA4 data flows into BigQuery - then it can be combined with other sources, such as Google Ads, Facebook, the store, a CRM or an ERP.

How long does implementing a single A/B test take?
The technical mechanism itself (variant assignment, DataLayer, GA4) totals from about 13 h. The time to analyze results depends on the scope and starts at 10 h.

Google Analytics 4 - how Hummerce sends e-commerce events to GA4.
Reports - BigQuery and Looker Studio - analyzing GA4 data in a broader context.
On hummerce.com: A/B tests (in Polish).

Want to see how A/B tests could work in your store? Contact the Hummerce team (in Polish).