10 condition types. AND / OR nesting. Reusable segments across experiments and flags.
What the browser already knows: location, device, URL.
Who the visitor is: cookies, return behavior, your own IDs.
Your app’s own attributes: anything you pass in.
Every condition is a leaf. Group leaves with AND or OR, then nest groups inside other groups. No artificial depth limit, no rules-engine vendor-speak, just the algebra your stats team already uses.
Conditions evaluate in the order that short-circuits fastest, so a 6-level rule tree costs the same as a flat one for most visitors.
AND
Location = US, CA
OR
Device = Mobile
AND
Browser = Chrome
New vs returning = returning
Custom attribute plan = 'pro'Save a rule tree as a segment. Reference that segment from as many experiments and feature flags as you want. Change the segment definition once, and every place it’s used updates.
Segments version automatically. Roll back the segment, roll back every audience that references it.
us-mobile-proThe browser snippet evaluates audience rules locally: the full datafile ships with the page, so there’s no round-trip. Context conditions (location, device, browser, URL) come from navigator, window, anddocument.
window.avsb.setAttribute.The Node SDK runs the same rule engine in your backend. Pass a user context object with any identity or custom attributes, and the SDK returns a consistent variation, no client round-trip needed.
Start free, no card. Build your first audience in minutes.