What Are Experiments? A/B Testing in Jurni

Understand how Jurni experiments assign Smart Link traffic between page variants, keep visitor assignments consistent, and compare performance without changing the campaign-facing URL.

Use a Jurni experiment when you want to send the same campaign audience to different page experiences and learn which approach performs better.

The basic model is:

Campaign / ad → Smart Link → experiment → control or variant → cart / checkout

The Smart Link is the URL you put in the campaign. The experiment decides which page that visitor receives.

When should I run an experiment?

Good reasons to test:

• a clear page hypothesis, such as a different hero message;
• a dedicated Jurni landing page vs. your current website page;
• proof placed earlier vs. later;
• shorter vs. longer education;
• a different campaign-specific page structure;
• a meaningful offer presentation when the underlying commercial offer is the same.

A test is less useful when the team has no idea what changed or why.

“New design vs. old design” can answer which whole page performed better, but it does not tell you which of 15 redesign decisions caused the difference.

For the marketer strategy behind a clean test, use Run your first A/B test.

The four parts of a Jurni experiment

1. Control

The current experience or existing page you want to compare against.

2. Variant

The alternative experience. Ideally, its difference from control maps to a clear hypothesis.

3. Smart Link

The campaign-facing entry URL. Traffic sent directly to a variant page does not automatically become experiment traffic.

4. Metric

The customer outcome you will use to guide the decision—such as CVR, RPS, AOV, or a relevant interaction rate.

Choose the primary metric before reading the result.

Example: test a stronger hero

Hypothesis:

We believe a benefit-led hero will improve purchase conversion for cold Meta visitors because the current feature-led headline does not explain why the product matters quickly enough.

Control:

• current hero.

Variant:

• benefit-led hero.

Keep fixed:

• product;
• offer;
• purchase options;
• CTA action;
• rest of page;
• campaign traffic.

Primary metric:

• CVR.

That is much easier to learn from than “AI made a better-looking page.”

What URL should go in Meta, TikTok, email, or creator traffic?

Use the Smart Link connected to the experiment when you want that campaign traffic assigned between variants.

Do not put the direct control or variant URL in the campaign and expect Jurni to split that traffic behind the scenes.

If you connect an existing Smart Link that already receives real traffic, starting the experiment can change where that existing traffic goes. Treat it as a live campaign decision.

Create an experiment →

How traffic assignment works

For a standard weighted test, Jurni assigns traffic according to the configured variant weights.

A few things to expect:

• a 50/50 setup does not mean every consecutive pair of visitors splits perfectly one-and-one;
• a returning visitor can retain the same assigned variant;
• repeatedly refreshing one browser is not a useful test of allocation;
• Routing Rules can intentionally send matching campaign parameters to a specific variant before the normal weighted split.

How traffic assignment works →

Experiments vs. page analytics

A Jurni experience can receive traffic outside an experiment—for example through a direct page URL, another campaign, or another path.

That means:

• experience-level sessions can include traffic that was never assigned by this experiment;
• experiment results should be read from the experiment context when making experiment decisions.

Why experiment and experience sessions can differ →

Testing against an existing website page

A variant can also point to a supported existing or external page when the tracking and attribution setup is correctly connected.

This lets you ask a useful business question:

“Should this campaign use our current Shopify / website page or the new Jurni experience?”

Because the two destinations can use different runtimes, carts, scripts, or tracking paths, QA both before interpreting the result.

How to test against an existing page →

How should I read the result?

Start with:

1. the date range for the actual test period;
2. your primary metric;
3. the observed difference between variants;
4. supporting journey/revenue metrics;
5. any traffic, page, product, offer, or tracking changes that complicate interpretation.

Do not declare a winner simply because one number is temporarily higher.

Read experiment results →

Then use the Optimization Loop to turn the result into a learning and follow-up hypothesis.

What happens when I pause or end a test?

Pause and End are traffic-routing decisions, not just reporting labels.

The current pause flow warns that traffic is sent to the shop home page. Ending asks you to choose the destination/winner that should continue receiving traffic.

Do not pause or end a live campaign without checking the exact routing consequence first.

Pause, end, and reuse an experiment →

Start here

Run your first A/B test

Learn the full marketer workflow from hypothesis to decision.

Create the experiment in Jurni

Follow the product steps for link, variants, weights, and launch.

Experiment prompt library

Use AI to create clean variants, QA them, and interpret results without overclaiming.

Troubleshoot experiment traffic

Check Smart Link use, sticky assignment, routing rules, and date ranges.