AI VISIBILITY AUDIT — SAMPLE

How AI engines answer your buyers’ questions —
and why Lumenkey isn’t in the answers

Prepared for: Lumenkey (fictional demonstration company) · B2B appointment-scheduling software
25 buyer-intent prompts × 4 AI engines · 100 recorded transcripts · Audit format v1.0

Confidential when prepared for a real client. In a real engagement every excerpt below links to a stored, timestamped transcript.

⚠ How to read this sample: in a paid audit, every number and quoted excerpt comes from real, stored transcripts of ChatGPT, Perplexity, Gemini, and Google AI Overviews responding to your buyers’ questions. Here, all of it is invented to demonstrate structure and depth.

1 · Executive summary

Buyers researching scheduling software ask AI engines an average of 6–10 evaluation questions before ever visiting a vendor site. Across the 25 highest-value questions for Lumenkey’s category, Lumenkey appears in 3 of 100 engine answers. Its two nearest competitors appear in 41 and 28. The causes are specific and fixable: Lumenkey is absent from 9 of the 11 third-party comparison pages the engines actually cite, its feature pages bury answers below marketing copy, and its site publishes no structured data. Section 5 lists 12 fixes in priority order; the first three are same-week wins.

Lumenkey mention rate (3/100 answers)

41%

“Calendia” (fictional competitor) mention rate

Structured-data types found on lumenkey.example

2/11

Cited comparison pages that include Lumenkey

2 · Citation scorecard

Engine	Lumenkey mentioned	Lumenkey cited (linked)	Calendia	SlotWise	BookNest
ChatGPT (search)	1 / 25	0 / 25	11 / 25	7 / 25	4 / 25
Perplexity	2 / 25	1 / 25	12 / 25	8 / 25	6 / 25
Gemini	0 / 25	0 / 25	9 / 25	6 / 25	3 / 25
Google AI Overviews	0 / 25	0 / 25	9 / 25	7 / 25	5 / 25

Method: 25 prompts, fresh sessions, no account history, engines queried on the same day. A real audit lists every prompt and stores every transcript.

3 · What the engines actually say (sample transcripts)

Prompt 4 — “best appointment scheduling software for multi-location clinics”

Fictional ChatGPT-style answer: “For multi-location clinics, Calendia is frequently recommended for its location-aware routing… SlotWise offers HIPAA-conscious intake forms…” Lumenkey: absent.

Root cause trace: the two sources cited by the engine are comparison listicles (fictional-saasreview.example, fictional-clinicstack.example). Lumenkey appears on neither, despite serving this segment.

Prompt 11 — “lumenkey vs calendia which is better for a dental group”

Fictional Perplexity-style answer: “Calendia publishes detailed per-location pricing… Information about Lumenkey’s multi-location support is limited; its website does not clearly state whether…” The engine says it can’t find Lumenkey’s answer — the page exists but the answer is in a paragraph 1,400 words down.

4 · Root-cause diagnosis

#	Cause	Evidence	Prompt clusters affected
1	Absent from the sources engines cite. 11 third-party pages were cited across the 100 answers; Lumenkey appears on 2.	Source list §A3 (real audit: URLs + screenshots)	“best X” (8 prompts), comparisons (5)
2	Answers buried on-page. Key feature pages open with slogans; direct answers appear 900–1,600 words in, inside stylized sections engines don’t extract.	Page analysis §A4	problem + evaluation prompts (9)
3	No structured data. Zero JSON-LD; no Organization, Service, or FAQ markup; no llms.txt.	Technical scan §A5	all — compounding factor
4	Entity inconsistency. Company described three different ways across site, LinkedIn, and directories; engines hedge (“appears to be…”).	Entity table §A6	brand + comparison prompts

5 · 30-day fix list (excerpt — 5 of 12 rows)

Priority	Action	Why	Effort	Owner
1	Get listed on the 4 highest-cited comparison pages that accept vendor submissions (named in full audit)	Cause #1 — engines answer from these pages, not from you	S	founder
2	Rewrite the multi-location feature page: 22-word direct answer first, self-contained sections, FAQ block	Cause #2 — engine said it “couldn’t find” an answer that exists	M	content
3	Add Organization + Service + FAQ JSON-LD sitewide; publish llms.txt	Cause #3	S	dev
4	Unify the company one-liner across site, LinkedIn, directories	Cause #4 — hedged descriptions	S	marketing
5	Publish a per-location pricing page (competitor’s citation magnet)	Scorecard §2 — pricing prompts are a 0/12 cluster	M	content

…rows 6–12 in the full audit, each tied to evidence, with a suggested 30-day sequence.

6 · What happens next

This audit is diagnosis. If you want execution: the Core package ($1,950) is rows 2–5 done for you — five pages rewritten answer-ready, paste-ready schema, entity checklist, and a re-test 14 days after you publish. Your audit fee credits in full. Or take the list to your own team — it’s written to be executable without us, and written Q&A is included either way.

Reminder: this document is a demonstration with fictional data. Lumenkey, Calendia, SlotWise, BookNest, and every transcript, number, and source above were invented to show the format of a real CiteLift audit.