A/B Testing

Test small changes against real orders

Describe a test in one sentence. ShopGuide builds it, checks it against your live storefront and splits your visitors between the versions. Then you compare them on real Shopify orders.

Start for Free See how it works
ShopGuide Experiments page with two running A/B tests, their variants, day counters and a stopped past experiment
🤔

You change things on a hunch

A new greeting, a hidden button, a different chat icon. Without a comparison, you can't tell whether a change helped or hurt.

🧰

A small test shouldn't need a project

Hiding one button or rewording one message rarely justifies a developer, a separate app and a tracking setup. So small ideas go untested.

🧾

Clicks aren't sales

A change that gets more clicks can still sell less. What matters is what visitors in each version actually ordered.

Meet A/B Testing

A fast way to test changes to your AI guide and your storefront against each other, with real Shopify orders as the scoreboard. In the app you'll find it under Extras, as Experiments.

💬

Describe it

Type the test in a sentence, like showing a pickup button or hiding it. ShopGuide builds the experiment. Prefer code? Write it as a snippet instead.

✅

Check it

ShopGuide reads your live storefront and checks the test actually changes something. Nothing goes live until you press Launch experiment.

📊

Compare it

Every variant gets its own results: users, orders, revenue, revenue per visitor and more, from real Shopify orders.

60-second overview

See a test go from one sentence to results

Describe a test in one sentence

Say what you changed and the versions you want to compare:

  • ShopGuide reads your live storefront and finds the element you mean
  • A preview shows each variant and what the visitor sees
  • If something is unclear it asks you one follow-up question
  • Prefer code? Write the experiment as a snippet in the editor

If an element can't be found on your live pages, the preview says so and the test can't be launched as it is.

Read the docs
Describe a test in one sentence
Test your chat and your theme

Test your chat and your theme

Each variant can change your AI guide or your storefront:

  • Chat bubble on or off
  • Proactive message on or off and its wording
  • Launcher icon and auto-scroll
  • Static or AI-written launch message
  • Show or hide theme elements like a pickup button
  • Restyle theme elements with CSS

Prices, products and checkout are out of scope. A/B Testing is built for small chat and storefront changes.

See what you can test

Results from real orders

Each variant gets its own row:

  • Users · Orders · Revenue
  • Revenue / User · Avg Order Value
  • Conversion Rate · Add to Cart Rate · Chat Interactions

Orders come from Shopify, in your store's currency. A ★ marks the variant leading on revenue per visitor.

Read about results
Results from real orders

The ground rules

21 days
minimum run, so you decide on a full dataset
2+
variants per test, split by the weights you choose
8
numbers per variant, from real Shopify orders
0
changes go live before you press Launch experiment

From a hunch to a decision

You added a local pickup button and you're not sure it helps. You type: “Make an A/B test for visible vs non-visible, 21 days.”

ShopGuide finds the button on your live pages, builds two variants and shows you a preview. You press Launch experiment. Visitors who allow analytics are split between the versions, and each keeps the same one on every visit from that browser.

After 21 days you open the results: orders, revenue and revenue per visitor for each version. You keep the one you prefer and stop the test. Its final numbers are saved.

ShopGuide doesn't pick a winner for you. You decide, with real orders in front of you.
How a test runs: describe it, check the preview, launch the experiment, visitors are split by your weights, compare variants on real orders, and you decide
How a test runs: describe, check, launch, split, compare, decide.

Questions? We've got answers.

What can I test? Can I test prices?
You can test your chat bubble on or off, the proactive message on or off and its wording, the launcher icon, auto-scroll, a static or AI-written launch message, and showing, hiding or restyling elements of your theme. Prices, products and checkout are out of scope.
How long does a test run?
At least 21 days. You set the end date when you create the test and can move it later, but not below 21 days. You can stop a test at any time, and its final results are saved.
Which visitors are included?
Only visitors who allow analytics. Without consent, a visitor sees your normal store and isn't counted. Known bots are left out of the results.
How do I pick a winner?
You decide. The results show each variant's orders, revenue and revenue per visitor, and a ★ marks the current leader on revenue per visitor. ShopGuide doesn't declare a winner or switch anything for you, so let the test run its full period, then keep the version you prefer in your settings or theme.

Test your next change before you keep it

A/B Testing is included in ShopGuide. Open Extras, choose A/B Testing and describe your first test.

Install ShopGuide See docs

Free until your chat has assisted 5 orders.

Recommendations
Product Details
Product Details