We use patented tech to capture over 1 million data points from the tests run by the world’s top brands.

We detect when brands run new tests, analyze them, and tag each one with metadata. Our analysts and AI use patterns across similar brands to provide guidance that wins 7X more often than traditional A/B testing.

Why it works

Every time a company we track runs a test, our engine gets smarter.

We’ve captured over 1M data points from real tests. Data is broken down by page type, brand, mobile/desktop, element, geography, and more. We run statistical analysis and train AI to continuously discover what actually works.

Real test from Airbnb. One of thousands used to train our AI.
Travel A/B test example, control versus variant with what was tested and the outcome

How we see A/B tests from any brand

Step 1

Our patented technology crawls the web to identify split tests based on public data

Step 2

It analyzes the tests it finds, and follows them to see what version the brand favors

Step 3

Our research team validates each test, adds metadata, and includes it in our database

For all stakeholders in the optimization process

Growth Teams

Home in on higher-converting website copy, layouts, and more by learning what wins for leading brands.

Book a demo →

Product Teams

Prioritize your backlog based on what is likely to win and use data to drive buy-in across teams.

Book a demo →

Designers

Get inspiration from what wins in tests to validate your assumptions and decisions as you design.

Book a demo →

Trends & strategy

Some of what we’re learning

Frequently Asked Questions

Where do you get the data?

Our patented engine uses public data to identify and analyze split tests. It is a lot like Google. It crawls the web and uses powerful and patented algorithms to make meaning from the data it finds. Our engine finds which tests are running, how many variants they have, and ultimately what wins and loses.

Does it matter what technology a company uses to run experiments?

Nope. We provide the insights on what to implement, and the customer can execute those recommendations wherever their site is managed.

How do you know what won?

We look at what version the brand favors and keeps live on their site. A team of analysts also reviews and validates every test before publishing it. Our analysts add considerable metadata about each test so we can run meta-analysis across tests to validate findings.

Do you have tests deep in registration and subscription flows?

Nope. We work only on public-facing pages that do not require a login.

Do I need to install a tracking pixel on my site to use you?

Nope. No pixel or tracking scripts are used.