Quickstart

MaxLift is warehouse-native A/B testing and feature flags. You assign users locally with a lightweight SDK, and MaxLift computes results inside your own warehouse — user-level data never leaves it. This guide takes you from sign-up to a running experiment.

1. Create your workspace

Sign up at maxlift.io and confirm the magic link. Your workspace is created automatically — no credit card.

2. Create an experiment from a template

Pick a template (e.g. Sticky Add-to-Cart) — it pre-fills a hypothesis, variants, and metric. Set a stopping rule (baseline conversion, minimum detectable lift, daily traffic) and MaxLift computes the required sample size and horizon so you don't stop early on noise.

3. Install the SDK

html
<script src="https://maxlift.io/api/sdk-ir" defer></script>

javascript
MaxLiftSDK.init({ tenantId: 'YOUR_WORKSPACE_ID' });

const userId = getYourStableUserId(); // the SAME id your warehouse events carry
const { variation } = MaxLiftSDK.experiment('sticky_atc_mobile', { userId });

if (variation === 'treatment') renderStickyBar();

// record a conversion when the goal completes:
MaxLiftSDK.track('sticky_atc_mobile', 'conversion', { userId });

4. Watch results

Exposures and conversions appear in your dashboard within seconds (no-warehouse mode) or after each warehouse refresh (warehouse-native). Before the planned horizon you'll see a 'collecting' state; at horizon you get a plain-English verdict with lift, confidence interval, p-value, and a sample-ratio-mismatch check.

The one rule that matters

The userId you pass the SDK must be the same identifier your warehouse events carry (e.g. your Segment anonymousId). That's the join key MaxLift uses to compute results in your warehouse.