That is not a guess. We run the entire category's purchase-intent question set and count how often each brand is mentioned, where it ranks, and what the assistants get wrong about it. First you get a map of the category — who wins, on which assistant, by how much — then we talk about fixing things.
One brand's own reading tells you exactly one thing: where it stands under that test surface, in those assistants' answers. What is actually useful is the distribution — every brand in the category, measured on the same question set, and how it moves over time.
Four separate things, not a bundle. Because we don't sell execution: measurement, monitoring and auditing are ours — how you fix it is your team's job, or your agency's.
You receive:
① The full AI visibility distribution of your category — not just you, but every brand in the field: who leads, who is slipping.
② Your share and rank change quarter over quarter; the trend starts speaking from period two.
③ Alerts: new entrants, share drops, newly introduced spec errors.
④ Every table carries its scope line (sample size × assistants × time window × retrieval on/off), so a third party can check it.
Public starting price ¥1,800/month · CNY; quarterly billing from ¥4,800/quarter, scaled by category size
Scope published in full · data is yours · deliverables carry a content fingerprint, independently verifiable
Request a benchmark sample and quote →Every month:
① Re-run of your category test set — trend dashboard of your AI Recommendation Share.
② Alerts: new entrants, share drops, newly introduced spec errors.
③ A prioritized visibility fix list: structured data, assistant crawler access, presence in the third-party sources assistants actually quote.
④ Quarterly review with a human, not a bot.
Public starting price ¥4,000/month · CNY; month-to-month, cancel anytime
The dashboard is yours · deliverables carry a content fingerprint, independently verifiable
See a sample dashboard →You receive:
① Your AI Recommendation Share across the assistant panel — mention rate, rank position, info accuracy, suppression ratio.
② A category-specific purchase-intent prompt set, with the answer log.
③ The error list: every spec, price or availability claim the assistants get wrong about you.
④ The competitor gap: who wins your customers' questions today, and on which assistant.
⑤ A prioritized fix list — 30 minutes walkthrough call included.
Public starting price ¥2,999 · CNY; scaled by category size, current quote in the one-pager
Delivery in 10 business days · prompt-level evidence pack · NDA available · for the whole field's distribution, see Category Benchmark
Request the one-pager →For two kinds of people:
① Brands that bought AI visibility work — your agency says visibility improved; we re-measure on a third-party scope and issue a fingerprinted audit report. Data, not talking points.
② The agencies themselves — "my client doesn't believe the result" is the biggest closing obstacle in this business. We issue the report; you may sign and republish it (figures unaltered).
③ Add a one-off snapshot attestation: a dated assistant answer to a dated prompt, plus a content fingerprint, from ¥199.
Public starting price ¥3,000/audit · CNY; quoted per project, annual terms for agency partners
We measure, we don't fix; we don't write articles and we don't place them — so there is no business of ours inside this report
Discuss an audit or partnership →AI recommendation logic is partly a black box and changes without notice. We commit to measured metrics — mention rate, position, accuracy — not to revenue outcomes.
Assistant knowledge updates lag the live web by weeks to months. Period one is always the baseline; trend statements start from period two or three.
We serve one client per brand per category. If your direct competitor is already a client in the same category, we tell you before you pay.
We run a limited number of free diagnostics for categories on our research roadmap — condition: results may be cited anonymously in published benchmarks. If your category is next, you see your data before anyone else does.
Or email hello@aibuy.org with “pilot” in the subject.