Skip to main content
Pre-release. This guide applies when the Integrations hub is enabled for your ABConvert account.
Connecting ABConvert to Segment sends one event to your Segment source each time a visitor is assigned to a test. Segment can route it to destinations whose credentials, mappings, consent settings, and enabled state accept the event. Use it to compare test groups in the tools you already operate.

What gets sent

ABConvert calls analytics.track with the event name and properties from Segment’s A/B testing spec. Destinations that understand the spec map these fields with no setup on your side. The event name is Experiment Viewed. The property names are Segment’s, not ours, because destinations match on them exactly: experiment_* fields describe the test and variation_* fields describe the test group.
The event fires once per visitor per test, on the page where the visitor is assigned. Returning visitors do not send it again. A visitor who lands in a different test group later sends a new event.

Setup

1

Install Segment on your storefront

Segment must already be loading on your storefront pages. Most Shopify stores use the Shopify by Littledata source, which adds Segment’s analytics.js to every storefront page. ABConvert does not install Segment.
2

Turn on the integration

In ABConvert, go to Integrations, find Segment under Data Platforms, and turn the switch on. No API key or write key is needed: ABConvert uses the Segment library already on your page.
3

Verify the connection

Open a page with a running test in a fresh browser, then open Segment → Sources → your source → Debugger. Experiment Viewed should appear within a few seconds. The Integrations page reads Called or queued after ABConvert successfully calls Segment’s browser library or adds the event to its queue. This confirms ABConvert handed the event to Segment’s library. It does not confirm Segment loaded, flushed, or delivered the event to a destination.

Using the event downstream

Destinations configured to accept the event can receive it. Confirm receipt in each destination; a source receipt does not prove every destination accepted it. A few common uses:
  • Amplitude or Mixpanel: filter any chart by experiment_name and break down by variation_name to compare test groups on your own events.
  • Warehouse: join experiment_id from the experiment_viewed table to your orders on the anonymous ID Segment attaches to every event.
  • Audiences: build an audience of users with Experiment Viewed where experiment_name equals your test, and variation_name equals one test group.
Use Segment when you want one source event to feed configured destinations. Use a direct Amplitude or Mixpanel integration when you need that vendor’s native exposure event.

Things to know

  • The event is sent from the visitor’s browser by the ABConvert script. Ad blockers or a declined consent prompt can stop it, so the count in Segment runs a little lower than the visitor count in ABConvert.
  • Segment’s library loads after the page starts rendering. ABConvert waits up to 10 seconds for the Segment client. If the client never becomes ready, the event is not replayed on a later page load.
  • Segment does not load inside Shopify checkout, so tests that only change checkout do not send this event.
  • Renaming a running test changes experiment_name for visitors assigned after the rename. Filter on experiment_id in downstream tools if you rename tests.
  • Test group names are the labels you set in ABConvert. If you renamed a test group to something like $49, that is what variation_name carries.
  • Use ABConvert’s own dashboard for statistical significance and revenue. Use Segment’s destinations for behavior across the rest of your data.