Platform dossier
Measure what AI tells your buyers,[1][1] Deterministic: scoring, ranking and trends are computed by our own code from stored runs. Models run prompts, classify sentiment and write prose. They never pick a number. See Methodology. then change it
Six connected surfaces: what the assistants answer, which questions are worth winning, who wins the ones you lose, which sources decide it, what to do next, and how to report it. Every number opens into the stored response behind it.
- Engines measured via provider APIs
- 6
- Buyer-intent categories
- 13
- Funnel stages
- 5
- Action stages, new to verified
- 7
Engines measured via provider APIs
Buyer-intent categories
Funnel stages
Action stages, new to verified
Nothing on this page is scored by a model. Every figure is deterministic code running over responses we stored.
01: AI visibility
Know which buyer questions mention you, and which ignore you
The Visibility Index is one number with three parts, and you can see all three. Mention rate carries 45%, position in the answer 25%, and citation ownership 30%. Change any input and you can see which part moved.
- Share of answer over time, so you can date a change rather than sense one
- Per engine, because ChatGPT and Perplexity frequently disagree about you
- Per market, kept separate and never averaged into a global figure
- Your top prompts, ranked by where you actually stand in them
Visibility index, out of 100
- Mention rate45%
- Position score25%
- Citation ownership30%
Computed by our own code from your stored runs. No model is ever asked to pick a score, so the same runs always produce the same number and you can check the arithmetic yourself.
02: Buyer prompts
Spend your effort on the questions that carry pipeline
Every tracked prompt gets a buyer-value score out of 100, built from ten weighted factors that sum to 100. High is 70 and above, medium is 45 and above. A comparison question and a definition question do not get the same budget.
Prompts are classified across 13 intent categories and 5 funnel stages, and each one reports an outcome: Win, Lost, Competitor owns, or Neutral. Where a signal is not connected, the factor falls back to neutral rather than being invented.
Outcomes tracked per prompt
- Win
- Lost
- Competitor owns
- Neutral
Prompt value score, out of 100
- Buyer intent18%
- Conversion potential12%
- Revenue relevance12%
- Competitor intensity12%
- Search demand12%
- Funnel weight12%
- Product relevance8%
- ICP fit8%
- Region importance3%
- Strategic importance3%
Weights sum to 100. Signals you have not connected fall back to neutral rather than being invented, so an unconnected workspace is scored conservatively instead of optimistically.
03: Competitor intelligence
See who replaces you in an answer, and why
When an assistant recommends someone else, the useful question is not who. It is what the model read to decide.
- A leaderboard of the brands actually named in your category's answers
- Head-to-head on any single prompt, run by run
- The verbatim claims made about a competitor, quoted from the response
- Source overlap: which domains cite them and not you
Best tools for enterprise onboarding
■ Omitted- Recommended instead
- Competitor A, named in 44 of 48 runs
- Why they win
- Cited from a comparison page on a review domain that never mentions you.
Illustrative example. Findings are generated from your own prompt runs.
04: Citations
Know which sources are shaping the answer
Every source an engine cites is stored at run time with its URL, the engine, its position and the capture time. Engines that return no sources are recorded as returning none. We never infer a source that was not given.
Every cited source, kept
Openable, with its URL, engine, position and capture time.
Source types, classified
Editorial, review, community or owned.
Your source mix
How far it leans on domains you do not control.
Top cited domains
In your category, ranked by frequency.
Citations over time
See a source arrive or disappear.
Opportunity gaps
Sources citing competitors and not you. Checked on request, never polled in the background.
05: Actions
Turn every gap into work someone owns
Each recommendation gets a priority score out of 100: business impact 35%, visibility opportunity 25%, competitive gap 20%, ease 10%, confidence 10%. The score is visible and so are its inputs, so you can disagree with the ranking on the record.
- New
- Accepted
- In progress
- Implemented
- Monitoring
- Re-scan needed
- Impact verified
That last stage is earned. An action reaches Impact verified only after a re-scan measures the prompt again and the numbers move. Shipping the work does not mark it done here.
Action priority score, out of 100
- Business impact35%
- Visibility opportunity25%
- Competitive gap20%
- Ease of implementation10%
- Confidence10%
Confidence is a term in the formula, which means a finding from a thin sample cannot outrank a well-evidenced one.
06: reports · paid feature
Report it without rebuilding it in a slide
Five report types, generated on demand over a 7, 30 or 90-day window, plus CSV export of the underlying rows. The Gap Sprint report packages the highest-priority open actions into something a team can work through. Agencies on Growth and above can put their own logo and accent colour on it and remove ours.
See which plans include themStated plainly: automated delivery on a schedule does not exist. You can save a frequency and a recipient list, but nothing in the product sends to them yet. When it does, this sentence changes.
07: Also in the platform
Four more surfaces, briefly
Sentiment and brand health: How assistants describe you, not just whether they name you.
Site intelligence: Audit scans across 25+ AI-readiness factors, with the specific fix for each.
AI traffic: Assistant referral traffic split into Observed, Estimated and Inferred, never summed.
Daily brief and alerts: An in-app brief and an email digest when something moves.
08: Questions
What people ask before they trust a number
Is the visibility score generated by AI?
No. Every score on this page is deterministic code running over stored responses. Models do three jobs here: run the prompt, classify sentiment and extract entities, and write prose summaries. A model never picks a number.
How are prompts prioritised?
By a buyer-value score out of 100 built from ten weighted factors, then cross-cut by 13 intent categories and 5 funnel stages. High is 70 and above. The factors are published on the methodology page.
What does displacement analysis show?
The prompts where a competitor is recommended and you are not, the brands named instead of you, what the answer claimed about them, and which sources cited them. It shows you the substitution and the evidence for it.
Can I read the answer a finding came from?
Yes. Every finding opens into the runs behind it. How many runs you can open depends on your plan: 1 on Free, 3 on Essentials, 10 on Professional, 25 on Growth, uncapped on Enterprise.
Do you show whether an action worked?
Only after a re-scan. An action can sit at Implemented indefinitely. It moves to Impact verified when the prompt is measured again and the result changes. We do not credit an action for a change we did not measure.
- The answer this page describes: what an assistant tells a buyer who asks. Sampled repeatedly across engines. See Methodology for the exact confidence formula.
Start with one prompt you think you already win.
Run it across the engines and see whether you do.