Measure · Fix Verification
Verified, partial, or failed.
Daily automatic re-scans of every query a published fix was meant to win, until the answer changes, or demonstrably doesn't.
The most dangerous moment in this work is right after you 'fix' something. You feel done. You move on. And the AI answer hasn't budged. You just don't know it yet.
Fix Verification keeps checking. It re-scans every query a fix was meant to win, daily, until the answer changes, or it tells you, honestly, that it didn't.
Declaring victory on day two is how brands lose quietly.
Before/after, engine by engine
The win/loss matrix shows mention and citation state per query per engine, before the fix and after. Green means the answer actually changed.
The lag is the truth
Engines re-read their sources on their own schedule; flips take days. Verification keeps checking, so you're not declaring victory on day two.
Failures escalate
A failed verification reopens the gap with what was learned attached: usually that a third-party source, not your page, is load-bearing.
FAQ
Questions you're about to ask.
How does verification work?
Daily automatic re-scans of the exact queries a published fix targeted. A win/loss matrix shows mention and citation state per query, per engine, before and after. Green means the answer genuinely changed, not that you hoped it would.
Why does it take days?
Because engines re-read their sources on their own schedule. A flip can take a week or more. Verification keeps watching across that lag so you're not fooled by a check you ran too early.
What are the possible verdicts?
Three, stated plainly: verified, partial, or failed. No grade inflation. A failed verification reopens the gap with what was learned attached: usually that a third-party source, not your page, is load-bearing.
What happens when a fix fails?
It doesn't disappear. The gap reopens with the new evidence, pointing you at the real lever, often a source you don't own, so your next attempt aims at the right thing.