The audit isn’t about how many tests you run. It’s about whether the results you act on are true, and whether the changes you make without a test are chosen by evidence or by whoever spoke last in the meeting.

Open your analytics, your testing tool if you have one, your store’s checkout settings, and your last ten tests or site changes. Score each check 0 to 2: 0 if it failed or nobody can answer it, 1 if partly true, 2 if clean. A tool’s “95% chance to win” doesn’t count as an answer to any of these.

A dashboard’s “chance to win” is a claim, not a check.

The twelve checks

1. You know each key page’s traffic and conversion rate · 4 minLook at: Four weeks of visitors and conversion rate for your home page, top product page and cart, by device.
   Good: Someone can produce the six numbers today, and knows which row of the traffic table each page sits in.
   Cost if wrong: You start tests that can’t finish, and believe the ones that stop early.
   Read next: How Much Traffic a Test Needs
2. Every test has a planned size before it starts · 3 minLook at: Your last five tests.
   Good: Each had a sample size and end date written down before launch, based on the smallest lift worth acting on.
   Cost if wrong: Tests end when someone gets bored or excited, which means they stop when the noise looks best.
   Read next: How Much Traffic a Test Needs
3. One primary metric, chosen in advance · 3 minLook at: The same five tests.
   Good: Each named one metric that would decide it, before it ran. Usually orders or contribution per visitor.
   Cost if wrong: With enough metrics, every test finds a winner somewhere.
   Read next: Why Winners Lie
4. Nobody stops a test early because it looks good · 3 minLook at: Whether any recent test ended before its planned size.
   Good: None did, or the tool uses a method built for checking results as they arrive, and the team knows which.
   Cost if wrong: Peeking can turn a 5% false-winner rate into 25% or more.
   Read next: Why Winners Lie
5. Winners are confirmed · 3 minLook at: What happened after your last three winners shipped.
   Good: Each was re-run, held back from a slice of traffic, or checked against the forecast after launch.
   Cost if wrong: You bank lifts that were never there.
   Read next: Why Winners Lie
6. The split is checked on every test · 3 minLook at: Visitors in each version of your last three tests.
   Good: Someone checked the split was as planned before reading the result.
   Cost if wrong: A broken split invalidates the result, and it happens more often than teams expect.
   Read next: Before You Believe a Result
7. An A/A test in the last year · 2 minLook at: Whether you’ve ever run two identical versions against each other on your testing tool.
   Good: Yes, and the split was even and no winner held up on a rerun.
   Cost if wrong: You’re trusting a setup nobody has checked.
   Read next: Before You Believe a Result
8. Research in the last quarter · 4 minLook at: Recordings or notes from customer sessions, and your post-purchase survey.
   Good: You watched at least five people use your site on a phone, and a live post-purchase question asks buyers what almost stopped them.
   Cost if wrong: Your test ideas come from opinions instead of from customers.
   Read next: Research That Finds the Leak
9. Known problems are fixed, not tested · 4 minLook at: Your list of known bugs, slow pages and errors.
   Good: There’s a list, it’s short, and nothing on it is waiting for a test.
   Cost if wrong: You spend traffic proving a broken thing is broken.
   Read next: Fix It, Don’t Test It
10. The product page answers the four questions · 5 minLook at: Your top product page, on a phone.
    Good: Total cost, delivery date, returns and reviews visible near the add-to-cart button, without scrolling far or opening anything.
    Cost if wrong: Shoppers find the answers at checkout, and about 70% of carts are abandoned before an order.
    Read next: The Product Page
11. Mobile has its own number and an owner · 3 minLook at: Your weekly report.
    Good: Mobile conversion rate is reported separately from desktop, and one person owns closing the gap.
    Cost if wrong: The device that brings most of your traffic gets the least attention.
    Read next: The Mobile Gap
12. A test log · 3 minLook at: Where past tests and site changes are recorded.
    Good: One searchable log, with the hypothesis, the planned size, the result and what changed because of it, for every test and every major change.
    Cost if wrong: You’ll run the same losing test again next year.
    Read next: The Brief and the Log

Score as you go; your band appears when all twelve are in.

Read your score

If you’ve never run a test, checks 2 to 7 score 0 and that’s fine: they’re about running tests well. Ignore the band and look at checks 1 and 8 to 12, which apply to every store. A clean 12 on those six is a strong start.