Ours Privacy Experimentation provides HIPAA-compliant A/B testing, redirect experiments, and personalization built into your CDP. No separate tool, no separate BAA.

BetaThis is in beta. The basics are stable, and more capabilities are on the way.

A/B Testing & Experimentation Overview

Ours Privacy Experimentation is HIPAA-compliant A/B testing and personalization built into your CDP. Run content experiments, redirect tests, and always-on personalization rules using the same visitors, audiences, and analytics you already have.

No separate tool. No separate BAA.

New: Manage experiments through Claude, ChatGPT, and other AI assistants. See Manage Experiments with AI.

Why Teams Choose Ours Privacy Experimentation

Drive experiments from your AI assistant. Create, edit, start, stop, and read live results from Claude, ChatGPT, or any MCP-compatible client — no dashboard required. See Manage Experiments with AI.
HIPAA-compliant by default. The platform is already HIPAA-compliant, so your experiments are too. No separate BAA or compliance add-on required.
CDP-native targeting. Experiments use the same visitor identity graph, audience segments, and event pipeline as the rest of your CDP. No integration required.
No visual flicker. Variants are assigned server-side before the page renders, so visitors never see the wrong content flash in before it's replaced.
Consent-gated. Only consenting visitors participate. Integrated with the Ours Privacy CMP at the pipeline level, not a client-side check that can fail silently.
First-party experiment data. Impressions and your existing conversion events flow into the same analytics pipeline you already use. No separate reporting silo.
Session replay and heatmaps per variant. See exactly why a variant won by drilling into replays and click maps for each variant.
Code-driven rendering when you need it. If your app owns the UI in React, Vue, or Next.js, use Headless Experiments and Headless Personalization to render variant-specific UI in your own code instead of relying on DOM mutations.

Experiment Types

Experiments come in four flavors. Pick by where the rendering happens:

Type	What It Does
Headless Experiments	Your application code reads the assignment from the SDK and renders the variant itself. Best for React, Vue, Next.js, and other framework-controlled UIs.
Server-Side Experiments	Your backend calls the assignment endpoint and renders the variant before the page reaches the browser. Best for SSR, edge functions, and non-browser clients. Browser and server converge on the same variant for the same visitor.
Content Experiments	The runtime modifies text, styles, images, or visibility on the same URL via DOM modifications. Best for static pages where the runtime can safely mutate the HTML.
Redirect Experiments	The browser redirects variant visitors to a completely different URL before the page renders. Best when you've built two distinct pages.

Personalization comes in two flavors:

Type	What It Does
Headless Personalization	Programmatic access to visitor signals — geo, UTM, device, and visitor status — via `window.ours_experiments.getVisitorContext()`.
Content Personalization	Always-on rules that show targeted content to a specific audience — no test, no winner.

All three experiment types support A/B/n with as many treatment variants as you need.

Key Features

Manage Experiments with AI: Run the full experiment lifecycle through Claude, ChatGPT, or other AI assistants over MCP
Installation: Load experimentation through the Web SDK, Tag Manager, or a manual script tag
Headless Experiments: Read assignments from the SDK and render variants in your own application code
Server-Side Experimentation: Call the assignment endpoint from your backend — SSR, edge functions, or non-browser clients — and render the assigned variant before the page reaches the browser
Content Experiments: Modify DOM elements per variant — text, styles, images, visibility, custom CSS/JS
Redirect Experiments: Send variant visitors to a different URL before the page renders
Headless Personalization: Access visitor signals from your own JavaScript for code-driven personalization and analytics enrichment
Content Personalization: Always-on content rules that target specific audiences
Goals & Conversion Tracking: Measure experiments against the events you already send your CDP, set a primary metric, and send new goal events
JavaScript SDK: Full reference for window.ours_experiments — assignments, events, and forced variants
Experiment Settings: Targeting, traffic allocation, visitor status, and consent

How It Works

Create an experiment in the Ours Privacy dashboard, via the REST API, or by asking your AI assistant
Define variants — modify elements visually with the Chrome extension, define DOM modifications or redirects via the API, or read assignments in your own application code
Set targeting — URL patterns, query parameters, visitor status, audience segments, or any combination
Start the experiment — the experiment script loads, window.ours_experiments.init({ visitorId }) runs, and targeting is evaluated on each page load
Measure results — pick a goal event you already send your CDP; Bayesian probability to be best updates as impressions and conversions come in
Declare a winner — stop the experiment and apply the winning variant permanently

Architecture

The experiment runtime loads as a lightweight JavaScript file from your custom domain. On each page load it:

Reads the visitor's existing variant assignment from a server-set cookie
If no assignment exists, deterministically hashes the visitor ID to a variant
Applies DOM modifications or redirects before the page fully renders, or exposes the assignment to your own code through window.ours_experiments
Tracks an $experiment_impression event via the CDP SDK (deduplicated per session)
Exposes window.ours_experiments, including explicit init(...), for access from your own JavaScript code

Getting Started

Install experimentation: If your site already runs the Ours Web SDK, you just add the experimentation token to your existing init call.
Create an experiment in the Ours Privacy dashboard
Define variants: Use the Chrome extension or the API to set up DOM modifications or redirect URLs
Start the experiment: Traffic allocation begins immediately
Track your goal event: Use your normal CDP event tracking, for example ours('track', 'signup_completed'). See Goals & Conversion Tracking for how experiments measure results.

Need Help?

Reach out to support@oursprivacy.com if you have questions about setting up experiments or interpreting results.

Overview