Map the path
- Write the path from ad click to record: the link, the landing page, the cart, the checkout, the confirmation page and the order in the back office.
- For each step, name the event or record that should fire and who checks it. Mark the key events the campaign is judged on, such as purchase.
Test with a real order
- Run the test from a clean browser session, such as a private window, and follow the path to a test order.
- Use a test product or test discount where the client allows it. Cancel or refund the order under the client's own process.
- Compare four readings: the conversions the ad account reports, the analytics tool, the order system and the payment record.
Made-up scenario: a test order worth 80 shows once in the order system and twice in analytics. The second row came from a confirmation page that fired again on refresh. It was fixed before launch, and the ad account was rechecked a day later because reporting can lag.
Check values and duplicates
- Check that the value sent with the purchase matches the order total the client counts, with the same tax and shipping rules. Write down which total it is.
- Check the currency and the decimal places. A value sent in cents where a report expects whole units makes every order look large.
- Check that each order sends one purchase with one order ID, so a refresh or a retry does not count twice.
Check consent and lost data
- Confirm what the site does when a visitor declines consent, and whether the campaign plan assumes data from those visits.
- Test with the usual privacy tools switched on. Write the share of data lost as a measured figure, or as not measured if nobody measured it.
Sign off and keep the evidence
- Save the test: date, device, order ID, the four readings and the result. Ask the client to sign the tracking sign-off before budget goes in.
- Set a first-day check that compares back-office orders with reported conversions, and write the gap down.
6 min read. Last updated 2026-10-07. Examples use made-up numbers.