Skip to content
×

A/B tests

Run two offers side by side and keep the one that earns more

Discounts & Bundles
Use with AI

Use with AI

Ready-made prompts Connect your AI to Pasilobus Install the CLI

This page

View as MarkdownPlain text of this page Open in ClaudeAsk about this page Open in ChatGPTAsk about this page

Ready-made prompts

Fill in the parts in angle brackets, then paste into any assistant.

Step by step on your store

Read https://help.pasilobus.com/discounts/ab-tests.md. Walk me through A/B tests in Discounts & Bundles for my store, step by step. My theme is <theme name> and I want <what you want to happen>.

Ask your own question

Read https://help.pasilobus.com/discounts/ab-tests.md. Answer this about Discounts & Bundles on my store: <your question>.

Connect your AI to Pasilobus Beta

Our MCP server lets Claude, ChatGPT, OpenClaw and other MCP clients use your Pasilobus apps. It is in beta and open to every Pasilobus account: you sign in with it when your assistant asks. Tools that change your store are marked as changes, so your assistant asks you first.

https://cloud.pasilobus.com/mcp

In Claude Code, add it, then run /mcp to sign in:

claude mcp add --transport http pasilobus https://cloud.pasilobus.com/mcp

In Claude on the web or desktop, go to Customize › Connectors › Add custom connector and paste the address. In ChatGPT, OpenClaw or another client, add it as a custom connector (MCP) with the same address.

Install the CLI

npm install -g pasilobus

Run your apps from the terminal or a coding agent. These only read, so they are safe to try:

pasilobus discounts discounts

Find discount offers by name, type or state

pasilobus discounts stats

Orders, discount given and order value from the store's offers

pasilobus apps lists your apps on the store; pasilobus <app> help lists what each app can do. How it works

On this page
  1. Set up a test
  2. What shoppers see
  3. Read the results
  4. End the test
  5. Good to know

An A/B test runs two offers of the same type at the same time. Each visitor is given A or B, sees only that offer, and gets only that offer's discount at checkout. When you have enough data, you keep the winner and the other offer goes to draft.

Set up a test

  • Open the offer you run now and click Duplicate. In the copy, change one thing: the discount, the tiers, the widget style or the badge. Changing one thing tells you what made the difference.
  • Go to A/B tests and click Create A/B test.
  • Pick your current offer as Offer A and the copy as Offer B. Both must be the same type, for example two quantity breaks.
  • Choose the share of visitors who see B. 50% gives the fastest result.
  • Click Start test. Both offers are switched on. If B was a draft, it now counts towards your plan's active offers.

An offer can be in one running test at a time.

What shoppers see

  • Each visitor gets a variant the first time they see one of the two offers, and keeps it on every page, on later visits and at checkout. The price they see is the price they pay.
  • A visitor who has no variant yet sees A. If they are given B on that first page, the page reloads once so it shows B.
  • Product page blocks, badges, bundles, gifts, free shipping and Build a box all follow the test.
  • Shoppers who clear their browser data may be given a new variant.

Read the results

Open a test to see each variant's numbers since the test started:

  • Visitors and Added to cart: shoppers who saw each offer and added it to the cart.
  • Orders and Revenue: paid orders from each variant's visitors.
  • Revenue per visitor and Conversion: the fairest comparison when the split is not 50/50.

The page says which variant is ahead. It waits for 100 visitors per variant and 10 orders before it shows a result, and it says winning once it is 95% confident the leader converts better. Until then it says ahead, so let the test run longer.

End the test

Click Keep A or Keep B under End test and pick a winner. The offer you keep goes on running for everyone. The other one is set to draft; you can switch it back on from the offer list. Ended tests stay in the list with their results. You can delete an ended test.

Good to know

  • The test needs the Discounts & Bundles app embed or an offer block on the page, which is where visitors are given their variant. A shopper who never sees an offer block gets A at checkout.
  • Run a test for at least one full week, so weekdays and weekends are both counted.
  • Offer performance under Reports still shows each offer on its own.

Support

Talk to a person.

The people who build our apps answer support themselves. Email us, or send a request from your Pasilobus account, and a person replies, usually within one business day. There's no chat.