Prompt Tracking
Prompt tracking is the practice of repeatedly asking a fixed set of buyer questions across AI engines and recording whether and how a business is mentioned each time.
It is the AEO equivalent of rank tracking, with one crucial difference: answers vary between users, sessions and phrasings, so a single observation proves very little. Only a consistent set of prompts checked repeatedly produces a trustworthy signal.
A sound prompt set mirrors how buyers actually ask — including comparative and constraint-laden questions — rather than the keywords a business wishes it ranked for.
Why it matters
Without it, AI visibility is anecdote. Someone tries one query, sees a competitor, and draws a conclusion from a sample of one. Tracking turns that into evidence.
An example
A recruitment agency writes thirty questions its clients actually ask — "best recruiter for finance roles in Bristol", "agency that places interim CFOs" — and runs them across four engines every week, recording whether the agency is named, in what position, and which sources were cited. After a month the trend means something; any single week does not.
Common mistakes
- Using the keywords you wish you ranked for instead of the questions buyers ask.
- Drawing conclusions from one run. Answer variability makes single observations unreliable.
- Changing the prompt set between runs, which makes the results impossible to compare.
Questions
- How many prompts should I track?
- Enough to cover the distinct questions your buyers ask — often twenty to fifty. More matters less than consistency: the same prompts, checked the same way, over time.
- Can I track prompts manually?
- Yes, and for a small set it is entirely workable. A spreadsheet recording the prompt, engine, date, whether you were named and your position is enough. Tools earn their cost when the prompt set or number of engines grows beyond what someone will check by hand.