Artixol
Use case ยท Lab

A fair A/B test on the same people.

Lab clones the same named personas into a control and a treatment. Same lives, clean memory, one difference. Then it shows both splits side by side.

In short

Artixol Lab runs an A/B experiment by cloning one audience into two identical samples with no study memory, running a different survey or message on each, and presenting the splits side by side. Because both groups are the same people, differences come from the treatment and not from who was sampled.

The question it answers

Which variant works better, with sampling noise removed?

How it works

  1. Pick the audience

    Use any sample you have already staffed.

  2. Write control and treatment

    Two survey versions or two message lines.

  3. Run on clones

    Each side gets the same people with clean memory. Leave the page; the run continues.

  4. Compare

    Open Past experiments for side-by-side splits.

What you get

The output

  • Control vs. treatment splits
  • Same people on both sides, so differences are the treatment
  • A history of past experiments
  • Export for the write-up
Example

Setup. 'Pay school fees in three taps' vs. 'Keep paying cash, wallets freeze when the lights go out' on 100 parents.

What you learn. The wallet line raised skepticism sharply in households with unreliable power, a fear the first line never touched.

Illustrative scenario, not a customer result.

When not to rely on it

Lab is the opposite of continuing a sample. If you want leftover memory, continue the audience instead of cloning it.

See the methodology for how to decide whether a result is strong enough to act on.

FAQ

A/B testing on the same people: common questions

Why clone instead of splitting the audience in half?

Splitting changes who is in each group. Cloning keeps everyone the same, so any difference comes from the treatment.

Can I run more than two variants?

Lab currently compares a control and a treatment. Run additional experiments against the same audience for more variants.

Bring a question. We will staff the audience.

Tell us who you need to understand. We will build the audience and run a study with you.