Website & branding · Landing page

A/B test your landing page headline

Run an A/B test on your landing page's headline or subheadline: visitors are automatically split between variants, and you see views, clicks, and click-through rate per variant before picking a winner.

How A/B testing works

A/B testing lets you try two or more versions of your hero headline (or subheadline) at the same time. Each visitor is automatically and consistently shown one variant — the same person always sees the same version, so the test stays fair. For every variant, the platform counts how many people saw it and how many clicked through toward signup, giving you a click-through rate to compare.

A/B testing is an optional feature that is switched on per account. If it isn't active for you yet, request it via the feedback option in your admin.

Managing experiments with your AI assistant

Experiments are created and managed through your AI assistant (named Benny by default — you can rename it) or via the API. Ask the assistant to create an experiment on the headline, give it at least two text variants (mark one as your current "control"), and tell it to start. Only one experiment can run per slot — one on the headline and one on the subheadline at a time.

While it runs, ask the assistant for results at any point: it reports each variant's views, clicks, and click-through rate. When you have a clear winner, stop the experiment and make the winning text your permanent headline in the Homepage editor or with the inline pen.

Step by step

  1. Make sure A/B testing is enabled on your account (request it via admin feedback if not).
  2. Open your AI assistant and ask it to create an A/B experiment on your landing page headline.
  3. Provide at least two headline variants, marking your current headline as the control.
  4. Ask the assistant to start the experiment.
  5. After enough traffic, ask for the results (views, clicks, click-through rate per variant).
  6. Stop the experiment and set the winning text as your permanent headline.

Good to know

  • Test one bold difference at a time — tiny wording tweaks need lots of traffic to show a signal.
  • Let the test run until each variant has meaningful views before declaring a winner.
  • Test the headline first; it has the biggest effect on whether visitors keep reading.

Related articles