Found before it launches, not in the first report.

• The pixel or tag fires on the actual thank-you page, not a page that also loads on refresh.
• The tag fires once per conversion, not once per page load or button click.
• Tested in an incognito window, not just the browser you're already logged into.
• The consent banner does not block the tag from firing for visitors who consent.

• Every platform is optimising toward the same definition of a conversion, not a softer engagement action.
• The value assigned to a conversion, if any, matches what the sale is actually worth.
• Duplicate conversions — a form fill and a purchase counted separately — are either intentional or excluded.
• The attribution window matches your actual sales cycle, not the platform's default.

• Every campaign, ad set and creative has a consistent UTM structure before launch, not added retroactively.
• The naming convention lets you tell channels apart in a blended report without opening each platform.
• Landing page URLs with UTMs have actually been clicked and checked, not just typed.

• Exclusion lists — existing customers, employees — are actually attached to the campaign, not just built somewhere.
• Retargeting audience sources are still populating, not silently broken by an earlier tag change.
• Consent and cookie settings match what the region you're targeting actually requires.

• Daily budget caps are set at the campaign level, not just assumed from the monthly total.
• Someone will check spend within 24 hours of launch, not at the end of the week.
• A stop-loss threshold — the number that triggers a pause — is agreed before launch, not decided in the moment.
• The bid strategy matches how much conversion data actually exists — automated bidding without enough signal usually spends before it learns.

Run this the day before launch, not the day of — most of these take a fix and a re-test, and neither is fun to do under a deadline you set yourself.