Seeds from product context, once
Generation is idempotent. If a library exists it is returned rather than rewritten, so a stray click cannot throw away questions you have curated.
Monitoring
The questions to track, generated from what you sell
50
questions per brand on Starter and Growth
fixedPromptsPerBrand in billing/fixed_plans.go. Generation seeds the first dozen; the rest are yours to write.
Picking prompts is where most AI-visibility tracking goes wrong. People track their brand name, watch it get mentioned, and conclude they are visible. Brand-name questions are the ones you already win. The questions that decide revenue are the ones where the buyer does not know you exist yet.
So the library is seeded from your product context: what you sell, to whom, and against which competitors. You can add and edit questions, and everything after that hangs off the list, since checks, corpus, leaderboard, and rank lookups all key on it.
Where it lives: Prompts, under Analytics in the dashboard, where you paste, suggest, pause and tag them. Regeneration is offered automatically after a repositioning.
What it looks like
IllustrativePrompt library50 of 50 questions
- 01best product analytics for small teamschecked 9 days ago · autoNext up
- 02alternatives to mixpanelchecked 7 days ago · auto
- 03analytics tool our CFO will approvechecked 4 days ago · you wrote this
- 04posthog vs amplitude for a seed startupchecked 2 days ago · you wrote this
- 05cheapest event analytics under 10k eventschecked today · auto
Longest unchecked goes next. A library twice as long degrades into slower coverage rather than into half of it never running.
What it does
Seeds a starter set, not the whole library
One generation returns about a dozen questions, which is short of the 50 a plan includes. That gap is on purpose. The questions worth checking are the ones you heard on a sales call, and a long generated list mostly buries them.
Regenerates when you reposition
Change the description, keywords, or competitors and the library is flagged stale. The next baseline rewrites it before picking a question, once per repositioning rather than on every run afterwards.
How it works
- 01Manual questions sit alongside generated ones and are never overwritten by a regeneration.
- 02Rotation is by staleness: the question checked longest ago goes next, so a big library degrades into slower coverage rather than into ignoring half of itself.
- 03Each question can be deactivated without deleting its history, so its past citations stay in the corpus.
What it does not do
- Generated questions are a starting point. Nobody knows your buyers' phrasing better than you do, and the good libraries we see are half hand-written.
- More questions is not better. Every question is asked on five surfaces, so padding the library with weak questions mostly adds rows you will not act on.
- We do not track question volume. There is no reliable public source for how often a phrasing is typed into an assistant, and inventing one would be worse than admitting that.
What you get
Starter and Growth include 50 prompts per brand. More is a custom quote.
Read the full plan comparison.
Live demo
See Prompt library on a real account, with no signup and nothing to configure.
Open the demoFrequently asked questions
Can I track competitor brand-name questions?
Yes, and it is one of the more useful things to add by hand. "Alternatives to <competitor>" tends to have the most citable corpus in a category.
How many questions can I track?
50 per brand on Starter and Growth. A custom quote can go from 25 to 300, and 25 is the floor because a smaller sample gives the share-of-voice number an error bar too wide to read. Automatic generation only fills the first dozen or so.
How do I know a question is worth tracking?
If the answer to it would name a vendor, it is worth tracking. If the answer would be a definition, it is a blog topic, not a visibility check.
Related features
Monitoring
Every feature we publish Prompt tracking
A row per prompt with rates over your chosen window, the latest verdict per engine, what changed since the last sweep, and a history page for each.
MonitoringAI visibility monitoring
Scheduled per-engine checks with a 0-100 score, tracked separately for being named and for being linked.
Citation corpusSource corpus
The accumulated set of pages the engines retrieved across every question and every run, deduplicated per URL and dated.
Run it on your own product
Start with the free visibility check to see whether the engines name you today, then run the scan that shows which pages they used instead.