How to track AI citations
Re-run the same questions on the same systems every month and record what comes back. What is being measured is an answer, not a page.
Tools exist that watch AI answers, and they are useful. What they cannot do is run your question set — the wording your buyers actually use, agreed with you, unchanged month after month. That is the comparison that tells you whether anything moved.
What this step answers
Is any of this working, and how would you know?
The tracked question set
The fixed list of buyer questions agreed at Baseline and re-run unchanged every month. Changing the wording restarts the measurement, so it does not change.
What we do
- Re-run the identical question set — same wording, same systems, same order
- Record every answer in full, not just whether you appeared
- Compare against last month and against month one
- Write down what we think moved it, attached to specific work
- Change next month’s plan accordingly. What moved gets more; what did nothing gets dropped
How to read it
Progress is questions entered, not a score going up. Entering three high-intent questions beats entering ten nobody buys from.
We report competitor movement beside yours, because a month where you gained nothing and the leader lost ground is not a bad month.
The flat months
Months one to three usually show no movement. You get those reports too, with what was published and why we still expect it to work.
Seeing them is the point.
What we do not promise
Not citations, rankings, traffic or leads. Nobody can guarantee those, including us.
AI systems are probabilistic and change continuously. The same question can return a different answer a week later with nothing having changed at either end. That is the reason for a fixed question set re-run on a schedule rather than a check made when someone remembers: one reading tells you almost nothing, and a run of them tells you most of what can be known.
What you receive
The Monthly Citation Report and a standing review.