Hypertune | Rapidly test thousands of website variations to discover hidden winners

Frequently Asked Questions

Can I set up tests myself?

Yes. For your simple ideas like changing text, images, fonts, sizes, colors, and hiding page sections, you can easily implement them yourself with our no-code visual editor.

For more complex changes like changing the page layout, your developer may need to implement them with custom JavaScript code.

Can you set up tests for me?

Yes. We offer a fully managed service where we'll audit your website, generate ideas, implement them and manage your campaign for you. Send us a message on the live chat to start the process.

What if I don't have ideas to test?

We offer a fully managed service where we'll audit your website, generate ideas, implement them and manage your campaign for you. Send us a message on the live chat to start the process.

How many ideas can I test at once?

Hypertune works best when you add as many variables as possible, each with 2 to 6 variations. The more ideas you add, the higher the chance of Hypertune finding a good combination of them.

How much website traffic do I need?

The more traffic you get, the faster Hypertune can work.

However, since Hypertune automates the testing process for you, and works faster than regular A/B testing, it works well even on very low traffic websites.

Typically, your website needs at least 2,000 monthly unique visitors to see results in a reasonable timeframe.

If you have a lower traffic website, you can use a higher funnel conversion goal like clicks or sign ups to get results significantly faster.

How does Hypertune work?

Suppose you have 5 versions of your title, 5 images and 5 backgrounds.

You should test all 125 combinations of them to maximize your conversion rate.

It's suboptimal to run one test for the titles, another test for the images and another for the backgrounds, as this doesn't account for the interaction effects between them — the winning title and winning image might not work well when put together.

A multivariate test solves this by equally testing all 125 combinations, but this requires too much traffic so takes too long.

Instead, our AI mimics evolution.

It starts by testing 5 random combinations out of the 125.

When it gets enough data on those, it kills off the worst ones and replaces them with new combinations similar to the best ones.

This process of "survival of the fittest" automates the testing process and can rapidly discover hidden winners even when there are thousands or millions of possible combinations.

Why does Hypertune work so much better than A/B testing?

There are 3 main reasons why Hypertune works so much better than A/B testing.

1) Since Hypertune explores so many variations, it can discover high performing ones that you wouldn't have even thought to test in a simple A/B test.

2) Since each variation is a combination of all your ideas, the variations can look very different from each other. This means Hypertune can test them faster — with less traffic — than if you tested your ideas in separate A/B tests.

3) Hypertune automatically discards bad variations and tries out promising new variations for you. So you don't have to spend time manually tracking experiment results and deciding what to test next. The AI will continue trying to improve the conversion rate until you decide to stop it.

What about statistical significance?

We've tuned our AI over millions of simulations to get you the maximum conversion rate uplift given your test setup.

Typically, the AI will statistically significantly beat your control and you can leave it running to continuously explore new variations.

You can also see the top variations discovered by the AI and run specific A/B tests to accurately measure their uplift.

Can I run regular A/B tests in Hypertune too?

Yes. You can run regular A/B and multivariate tests in Hypertune too.

Does Hypertune support segmentation and personalization?

Yes. Hypertune lets you target your campaigns to specific audiences, so you can discover the best variations for each user segment.

How is this different to a multi-armed bandit?

A multi-armed bandit is similar to a regular A/B test in that it typically only tests a few variations.

The difference is that a multi-armed bandit shifts traffic to high performing variations as the test progresses.

However, it's still not feasible to explore thousands of variations with a standard multi-armed bandit.

To achieve this, Hypertune leverages a type of AI called evolutionary optimization. This starts by testing a few random variations. When it gets enough data on those, it kills off the worst ones and replaces them with new variations similar to the best ones.

This process of "survival of the fittest" automates the testing process and can rapidly discover hidden winners even when there are thousands or millions of possible variations.

This is not possible with standard multi-armed bandits or A/B tests.

Rapidly test thousands of website variations to discover hidden winners

Discover hidden winners

Automate your testing

Get results fast

Features

Pricing

Frequently Asked Questions