Writing
What we have learned building this, including the awkward parts.
The mechanics of testing a Shopify store: the arithmetic, the tradeoffs, and the things a pricing page has no room to say. Most posts have figures you can drag, switch and break.
Latest
5 October 2026 · 5 minOrder tags and analytics apps: your test groups, in Triple Whale and GA4
Settings now has Integrations: tag every tested order with its group, and send each exposure to Triple Whale, GA4 and Shopify customer events. Both opt-in. What is sent, in what shape, and how to read it.
Read the post- 01The tags and metafield on one ordertype a test and a group
- 02Every place the group goespick a channel
- 03Does this page view send an eventflip the conditions
Start here
The arithmetic every test is subject to, whichever tool runs it.
- Updated 5 October 2026 · 7 min
Anti-flicker: what it costs, and when hiding the page is wrong
Client-side A/B testing has to hide the page for a moment or show the wrong version. Both are bad. Here is the tradeoff, the rules we hold ourselves to, and what each page load actually does.
3 interactive figures- 01The same page load, with and without the overlay
- 02Seven page loads on one timeline
- 03What a page view downloads
- Updated 5 October 2026 · 8 min
How many visitors does a Shopify A/B test actually need?
The honest number, worked out from the arithmetic rather than a rule of thumb, and what it means for a store doing under 10,000 visitors a month. With a calculator.
3 interactive figures- 01Halve the lift, quadruple the traffic
- 02How long your test needs
- 03Peeking at a test with nothing to find
- Updated 5 October 2026 · 8 min
Measure revenue, not clicks, and what that costs you
Click-through rate is easier to move, easier to prove, and frequently wrong. Why we judge tests on revenue per visitor instead, and the two things it makes harder.
Read the post3 interactive figures- 01One experiment, three verdictsswitch the metric
- 02One big orderdrag the order value
- 03Following an order to its groupbreak a link
Guides, one per test type
How each kind of test works on a Shopify store, and what to check before starting one.
- Updated 5 October 2026 · 13 min
How an experiment is set up: every tab, and what each option does
Six tabs, one draft, one save. What Test groups, Modifications, Targeting, Measurement, Preview and Results each decide — and the options on them that change your answer rather than your setup.
3 interactive figures- 01Six tabs, and what saving does
- 02Where a visitor lands
- 03One visit against the audience
- Updated 5 October 2026 · 8 min
Running a content test: headlines, buttons and the selector problem
The cheapest test to set up and the easiest to break. How to pick elements that survive a theme update, and why a selector that matches nothing is worse than an error.
2 interactive figures- 01The four operations
- 02Which selectors survive
- Updated 5 October 2026 · 7 min
Running a redirect test: two pages, one question
Split-URL testing on Shopify. Why the origin URL is the whole setup, what happens to your ad attribution, and the one mistake that would redirect your entire store.
2 interactive figures- 01A shopper's visits to the origin
- 02A content test under a redirect
- Updated 5 October 2026 · 7 min
Running a styles and JavaScript test: power, and the guardrails
Inject scoped CSS and JS per group on the theme you already have. Nothing to set up in Shopify, and the one test type that can genuinely break a storefront.
2 interactive figures- 01One stylesheet per group
- 02When the change lands
- Updated 5 October 2026 · 8 min
Running a theme test: the heaviest change you can make
Split traffic across two whole themes. The highest-value test and the highest-risk one — including the mistake that makes a control collect nothing at all.
3 interactive figures- 01Can this theme test start?
- 02Counting loaders on a page
- 03What the browser does
- Updated 5 October 2026 · 7 min
Running a template test: two page layouts, one product
Compare alternate Online Store 2.0 templates for a page type. Narrower than a theme test, so it reaches an answer sooner — and it leaves you something you can publish.
2 interactive figures- 01Can this comparison run?
- 02A shared link, opened by someone else
- Updated 5 October 2026 · 8 min
Running an offer test: does the discount earn what it gives away?
Compare 10% off against free shipping against nothing at all, applied at the cart by Shopify rather than in the browser. The setup, the arithmetic, and how to read the answer.
3 interactive figures- 01What each offer has to earn
- 02One shopper in the 10%-off arm
- 03Which metric the results page will judge
- Updated 5 October 2026 · 13 min
Running a price test: the mechanism, the checkout, and what to believe
Charge different shoppers different prices for the same product, up or down, on any Shopify plan. How the charge is kept equal to the page, what the shopper sees at checkout, and how to read the answer.
3 interactive figures- 01Control against one test price
- 02How your test will charge
- 03One test price, every market
Release notes
What changed in Zinx Signal, and why.