Monitor what AI answers say about your competitors
Your competitor just got named “the best option for enterprise teams” by ChatGPT. You’d want to know — and more importantly, you’d want to know which prompts they’re winning and where they’re absent.
The competitive prompt set
Build the set around buying moments, not your brand:
- Category prompts: “best
tools”, “top for ” - Comparison prompts: “X vs Y”, “alternatives to X” — including yourself as X
- Problem prompts: “how to
” — who gets recommended as the solution - Displacement prompts: “is X worth it”, “X pricing too high” — where sentiment tilts
Run it across engines and the markets you compete in. entities[] gives you the named brands per answer; sources[] shows whose content is feeding the recommendation.
What to extract per run
| Signal | Field | Question it answers |
|---|---|---|
| Share of recommendations | entities[] | Who’s named, and in what order? |
| Evidence sources | sources[] | Whose content is winning citations? |
| Fan-out overlap | searchQueries[] | Which searches feed this answer — do you rank there? |
| Commerce presence | shoppingCards[] | Whose products appear with prices? |
The cadence that matters
Weekly for the core set, daily during launches or pricing changes. The alert that matters isn’t “competitor mentioned” — it’s a transition: they entered a prompt’s answer where they weren’t before, or their citation displaced yours. Diff runs, alert on state changes.
Reading the counterplay
When a competitor starts winning a prompt, sources[] tells you why: which page of theirs got cited. That page is the brief — the format, the claim, the depth that the engine rewarded. Match the format, make a better claim, measure whether the citation moves.
Set it up with an async batch — prompt × market × engine matrix, webhook delivery, diff in your own storage. The share-of-voice framework turns the raw data into reportable numbers.