Getting started
Introduction
What freeabtest.com does, how the pieces fit together, and the terms used throughout these docs.
freeabtest.com is an A/B testing platform. You create an experiment with two or more variations, add a small JavaScript snippet to your site, and the platform splits visitors between the variations. It records who saw what and who converted, and tells you whether the difference is real or likely noise.
These docs describe what the product does today. Features that are planned but not built yet are marked Coming later. See the roadmap for the bigger picture.
New here?
Follow the Quick start. It takes you from sign in to reading your first result in about fifteen minutes.
How it works#
- 1
Assign
When a page loads, the SDK calls
POST /api/assignwith the experiment ID and a visitor ID. The server picks a variation and stores the assignment. - 2
Apply
Your code receives the variation name (and an optional redirect URL) and changes the page to match.
- 3
Track
The SDK records a page view automatically. You call
ABTest.trackConversion()when the visitor does the thing you care about. - 4
Analyze
The results page compares each variation with the control: conversion rate, lift, a 95% interval and a p-value.
Core concepts#
| Term | Meaning |
|---|---|
| Organization | Your account. Created automatically the first time you sign in. Plans, limits and suspension apply at this level. |
| Project | A site or app you test on. Every organization starts with one project called Default project. |
| Environment | Each project has three: Development (dev), Staging (staging) and Production (prod). Data from each is kept apart. |
| API key | SDK keys (cro_pk_...) are public and used in browsers. Secret keys (cro_sk_...) are private and used from servers. |
| Experiment | A test with a name, a hypothesis, a traffic allocation and a control plus 1 variation on free, or up to 50 on paid plans. |
| Variation | One version of the page. Exactly one variation is the control (your current version). |
| Visitor | Identified by a visitor ID. The SDK creates one and stores it in localStorage, or you can pass your own. |
| Assignment | The record that a visitor belongs to a variation. It never changes for the life of the experiment. |
| Conversion | An event you send when a visitor reaches your goal, such as a signup or a purchase. |
What is available today#
- URL split tests with a control page and variation pages (1 variation on free, up to 50 on paid).
- Automatic redirect from the control page to the assigned variation, without a flash of the wrong page.
- Sticky, deterministic assignment based on a hash of the visitor and experiment IDs.
- Click goals by CSS selector, with a primary goal and deduped conversions per visitor.
- Results with conversion rate, lift, Wilson 95% intervals and a two-proportion z-test.
- Separate dev, staging and prod environments, each with its own keys.
- An audit log of every change to experiments and keys.
What is coming later#
Coming later
Session recordings, funnels and a visual editor are on the roadmap. Heatmaps are live on paid plans.