01
Input
Content roadmap
Gaps become briefs. The questions where a competitor is named and you are not are the highest-value pages you could write.
Feeds directly into Content Systems.
AI Visibility
Measure whether the machines say your name
You cannot manage what nobody is reading. Every night we run your buyers’ real questions across five answer engines, capture what comes back verbatim, and show you exactly where you are named, cited, or absent.
412
prompts per client in a typical panel, sampled nightly
5
engines: AI Overviews, ChatGPT, Perplexity, Gemini, Claude
Verbatim
answers captured, not a score you have to trust
Why a score is not enough
I
The wording is the finding
Knowing you appear in 42% of answers tells you nothing actionable. Reading the sentence, how you are described, what you are compared to, which competitor is named first, tells you what to fix.
II
Answers drift without an announcement
A model update or a competitor publication can flip a question overnight. Monthly sampling finds out four weeks late; by then the answer has hardened.
III
Named is not the same as mentioned
Being listed among eight options is not visibility. We grade every reading as NAMED, CITED, MENTIONED or GAP, because the difference is the difference between a shortlist and a footnote.
The Instrument
No off-the-shelf tool reads answers the way we need, so we built the harness ourselves. It is the same one behind every claim on this site.
We also run Peec AI, Search Console, GA4 and Ahrefs alongside it, but the nightly panel is ours, and the historical series it produces is yours.
In-house harness
Historical series
Exportable raw data
What a reading contains
Four layers, delivered nightly and summarised monthly.
Every prompt, every engine, graded consistently, with the verbatim answer stored alongside it. You can read the actual sentence that produced the grade, which means nobody has to argue about methodology.
Consistent grading
Verbatim stored
Per-engine breakdown
The percentage of your priority questions where at least one engine names you. Tracked as a time series so you can see the effect of a publication two weeks after it ships, not two quarters later.
Time series
Per-question detail
Publication attribution
For every gap, we record who is being named instead and what they published to earn it. Most of our clients’ best content decisions have come out of this column rather than any keyword tool.
Competitor named
Source they cited
Gap-to-brief handoff
Models regularly describe companies with outdated pricing, retired products or plain errors. We flag factual drift so it can be corrected at the source before it propagates into every answer in your category.
Factual drift alerts
Sentiment grading
Correction routing
What you do with it
The reading is only useful if it drives the next sprint. Four ways ours does.
02
Alert
Defence triggers
A position lost overnight opens a ticket the same morning, with the answer text and the likely cause attached.
Usually a competitor publication.
03
Proof
Board-ready reporting
One page: Share of Answer, movement, what caused it, what is next. Understandable by people who do not do this for a living.
Monthly.
04
Product
Category intelligence
A year of nightly readings is a dataset about your market that nobody else holds, including your competitors.
Often becomes publishable research.
Getting started
- 01 Prompt panel design The real questions, drawn from search data, sales calls and support. Usually 200-500 to start. Week 1
- 02 Baseline reading The first full sample across five engines. This is what everything is measured against. Week 1
- 03 Nightly cadence The harness runs. You get access to the raw readings, not just a dashboard. From Week 2
- 04 Gap-to-sprint loop Gaps become briefs, briefs become pages, pages get re-read within a fortnight. Ongoing
- 05 Quarterly panel review Questions change as your market does. We prune and extend the panel every quarter. Quarterly
Next
Get one week of readings, free
Send us your ten most valuable buyer questions. We sample them nightly for a week across five engines and hand you the raw answers and the chart. No contract, no obligation.