Before starting, make sure Setting up price testing is done for your theme — in particular that price elements are tagged. A test on an untagged theme publishes fine and changes nothing.
Step 1 — Start the test
In the Cooee dashboard go to Experiments → Create and choose Price test. You will land on the experiment form with a control arm and one variant already created.
Step 2 — Choose your audience
For everyone, leave the audience filters empty. For a specific slice of traffic, add filters on any of:
landing page URL
landing page path
UTM source, medium or campaign
Filters match against the page the visitor entered the site on, not the page they are currently viewing. A visitor who arrives on a campaign landing page and then browses elsewhere remains in the test for that visit.
Keep visitors in the test once they qualify is off by default. A visitor is in the test only while they still match your filters, so someone who arrives from a campaign today and returns directly tomorrow sees normal prices on the second visit. Turn it on to keep them on the test price for the rest of the test, however they return.
Leave off for a clean read on one campaign's traffic
Turn on when you care about the whole purchase journey, including people who come back later to buy
Step 3 — Write your hypothesis
State the change and the outcome you expect. This is not bureaucracy — it is what stops a test being reinterpreted afterwards to fit whatever the numbers happened to do.
A usable hypothesis names the change, the expected direction, and the metric:
Raising the price of our top three moisturisers by 15% will reduce conversion by less than 10%, increasing revenue per visitor.
Step 4 — Set the traffic split
Use the slider to divide traffic between the arms. Anything from 1% to 99% is allowed.
50/50 reaches a conclusion fastest
Weighted towards control (say 80/20) limits exposure if you are nervous, at the cost of taking longer
Step 5 — Choose products and set prices
Add the products to test, then set the price change for each on each variant arm. For every product you set a direction (increase or decrease), a type (percentage or fixed amount) and a value.
The change is an adjustment, not an absolute price. A 10% decrease applies to whatever the product's price is at the time, so a product whose base price changes mid-test keeps a sensible test price.
The control arm has no rules. It is your existing price, untouched.
Use percentage adjustments on multi-currency stores. A fixed amount carries no currency, so "200" is applied as 200 of whatever currency the visitor sees.
Compare-at price
You can also adjust the compare-at price on an arm, which changes the struck-through figure shown next to the price. Use it when you want to test the apparent saving rather than only the price paid.
If you leave it empty, Cooee handles the labels automatically: when the test price is lower than the original, the original is shown struck through with a savings label. A product that already had a compare-at price keeps it, as long as it is still above the test price.
Raise Shopify base price
Intended for increase tests — raise the product's Shopify price to the higher test price and serve the original price as the variation, so external feeds show the higher list price. This is not functional end to end yet, so leave it off.
Step 6 — Review before you set it live
Every product has a price change on the variant arm — a product with no change is measured but does nothing
The test name says what the test is, so it is recognisable in a list six weeks from now
The hypothesis is filled in
The audience is what you meant
No product here is in another active price test — only one adjustment is ever applied, and both tests become unreadable
Tips for your first test
Start with one to three high-traffic products. A test spread across twenty low-traffic products will not conclude.
Use a meaningful change. A 2% move needs far more traffic to detect than a 15% one.
Plan for two to four weeks, enough to cross weekly buying patterns.
Change one thing. Testing a price change and a page redesign at once tells you nothing about either.
Do not set it live yet — go to QA checklist for a price test.