My citations went down this month
Why this happens, when it is normal variation, and the specific signs that it is a real decline worth acting on.
Most single month drops are normal variation rather than a decline. Here is how to tell which one you are looking at.
Why the number moves on its own
Citation share is a sampled estimate, not a count. Answer engines do not return identical answers to identical questions, so the same measurement run twice produces slightly different results.
The dashboard does not show a metric called citation share. To check for a drop, use Mentioned or cited in on Home, Cited in a Query's engine answers, and Discovery on Competitors. Discovery can fall between weeks.
That variation is built into the metric. It means a certain amount of month to month movement in both directions carries no information at all.
| What you see | Likely reading |
|---|---|
| A small drop, trend still rising | Noise. Ignore it |
| Down for one month after several up | Likely noise |
| Down three months running | Possibly real. Worth investigating |
| Down sharply across every engine at once | Worth checking the setup, not the content |
| Down on one cluster of questions specifically | Real, and the most actionable kind |
The real causes, when it is real
A competitor published something better
The most common genuine cause. Someone built better coverage on questions you were winning, and the engines now prefer them.
How to spot it: a specific rival improves on the Competitors screen, and the engine answers for the Queries you lost name them.
What to do: tell us which questions matter most. This usually calls for a content response, although we cannot promise the answers will change.
An engine changed how it builds answers
These systems get updated. A change in how one of them weighs sources or formats answers can move everybody's numbers at once.
How to spot it: a drop concentrated in one engine while the others hold, affecting many questions at once rather than a cluster.
What to do: mostly wait. Reacting to an engine change inside a single month usually means chasing something that settles on its own.
Your site changed
A redesign, a migration, a CMS change, a new security layer, or content removed in a cleanup. Any of these can reduce what is readable.
How to spot it: the drop starts in the same month as a change on your side.
What to do: tell us what changed. This helps us check whether the change reduced what answer engines can read.
The question set changed
If Queries were added, your numbers are now measured against a larger set. Adding twenty Queries you have no coverage on yet can lower the proportion on Home even though nothing got worse.
How to spot it: Y in Mentioned or cited in X of Y active queries has changed.
What to do: nothing. The denominator moved, not your performance. Worth asking us to confirm.
Something broke in the setup
Worth ruling out if the drop is sharp and across the board.
How to spot it: Requests over time for AI crawlers also dropped on the Traffic screen.
What to do: contact us. That combination is the one pattern that warrants a same week conversation.
What to actually do about it
To look back, open a question on Queries and choose an earlier week, or choose a week on Competitors.
Compare with your earlier numbers, leaving out the last month
If the longer trend is still rising, you are looking at noise.
Check whether AI crawler requests moved too
On the Traffic screen, citations down with requests steady points to content. Both down together points to setup.
Check whether anything changed on your side
Deploys, migrations, redesigns, security changes, content deletions.
Look at which questions specifically
A cluster is a brief. A scatter is noise.
If it has been down three months running, ask us directly for an honest read on whether the approach is working for your category. That is a fair question and the answer should not be a chart.