Decision-Stage AI Visibility: The Engines Buyers Ask
Buyers ask AI which option to choose before they reach your site. Decision-stage AI visibility is whether you win that recommendation, not just appear.
Jul 12, 2026•21 min read
Sample report
The whole audit, not a screenshot of it.
Below is a complete Radar audit exactly as a client receives it: all 13 tool scores, what each of the four engines actually said, the competitive benchmark, five findings with evidence, and a dated fix list. Nothing is trimmed to make the product look better. The subject scored 33 out of 100.
The client is de-identified. Their name, domain, the same-name businesses they collide with, and the people named in the engine responses are all removed. Every score, count, and finding is unedited.
Pixelmojo Radar - 13-tool audit
AI Visibility Audit
- Subject
- Boutique travel agency (client de-identified)
- Period
- First audit baseline, July 2026
- Engines
- ChatGPT, Claude, Perplexity, Gemini (live), Grok report-only
- Method
- How Radar scores this
Overall visibility
33
out of 100
Grade D
Executive summary
AI engines can reach the site but cannot identify the company, and when they try, they often describe a different business with the same name. That is fixable.
- Overall AI visibility: 33 out of 100 (grade D). The technical foundation is solid. The losses concentrate in three structural gaps that score zero: structured data, llms.txt, and AI share of voice.
- Top finding: identity confusion. Three of four AI engines described a different same-named business when asked about this company. Until that is corrected, every other improvement leaks value to companies that share the name.
- Absent from buying conversations. When AI engines answer "best travel agencies" style questions, this company appears 0 times out of 8 competitive queries. Share of voice is 0 percent, rank 119 of 118 brands found.
- Top action: publish an identity layer this week. Organization schema with a disambiguating description, plus an llms.txt file. Both are hours of work, not a rebuild, and they attack the two zero scores and the confusion problem at once.
Overall visibility score: 33/100 (D)
In plain English: AI engines can crawl the site freely, and about half the time they say something generic about the company when asked directly. But they cannot verify who it is, they confuse it with at least three other businesses, and they never recommend it in the category questions where buyers actually are.
Grade scaleA to F
Tool breakdown
All 13 tools, sorted by score. Each one measures a separate, independently verifiable thing, which is what makes the composite defensible rather than opaque.
| Tool | Score | Grade | What it measures |
|---|---|---|---|
| Hallucination Check | 70 | B | Are AI statements about you accurate |
| Reddit Presence | 55 | C | Community mentions AI trains on |
| Answer Engine | 49 | D | Do engines cite your URL when asked about you |
| Citation Check | 49 | D | How often engines mention you across query types |
| Crawl Check | 44 | D | Can AI bots access and read the site |
| Robots Analyzer | 42 | D | robots.txt configuration for AI crawlers |
| AI Readiness | 41 | D | Composite technical readiness |
| Source Influence | 39 | D | Does your domain appear in engines' source sets |
| AEO Auditor | 25 | F | Is content structured for AI answers |
| Brand Disambiguation | 20 | F | Do engines know which same-named company you are |
| llms.txt | 0 | F | AI site-guidance file (missing, 404) |
| Schema Audit | 0 | F | Structured data (0 blocks across 5 pages checked) |
| Prompt Share of Voice | 0 | F | Presence in category buying questions |
LLM breakdown
What each engine returned when asked about the company directly. These are live queries, not a cached snapshot.
| Engine | Knows you? | Cites your URL? | Notable response |
|---|---|---|---|
| Perplexity | Partially, blended with others | Yes | Described the company correctly as custom-itinerary planning, but presented two same-named agencies in other countries as equally likely matches, and reported a founding year the site contradicts. |
| Gemini | Confused | No | Led with a same-named LLC registered in 2024 by two named founders, an entirely different entity profile than the site presents. |
| ChatGPT | Generic only | No | Produced a plausible generic travel-agency description with no specific knowledge. Absent from all competitive answers. |
| Claude | No | No | Stated it had no reliable specific information about a company by this name, and speculated it might be a blog or a regional business. |
| Grok (report-only) | No | No | Stated it had no information on a company or service by this name. |
Mention rates when asked about the brand or its category: ChatGPT, Perplexity, and Gemini mentioned the brand in 2 of 4 queries each. Claude and Grok in 1 of 4.
Query coverage
| Query type | Expected | Actual | Severity |
|---|---|---|---|
| Brand query ("What is [company]?") | Accurate description, URL cited | Generic or confused description; URL cited by Perplexity only | High |
| Competitive queries ("best travel agencies", 8 total) | Appear in at least some answers | 0 mentions out of 8 | High |
| Category source sets (sites engines cite for travel) | Domain present | Domain absent from every provider's source set | High |
Competitor benchmarking
AI share of voice across "travel agencies" prompts.
| Brand | Share of voice | Mentions | Engines citing them |
|---|---|---|---|
| booking.com | 6% | 8 | all 4 |
| kayak | 5% | 7 | 3 |
| expedia | 4% | 6 | 3 |
| This client | 0% | 0 | 0 |
Rank: 119 of 118 brands found across 5 prompts and 4 providers. The realistic goal is not outranking booking.com on head terms. It is appearing at all in long-tail questions that match the actual specialties.
Same-name interference
15 same-name competitor URLs across three countries absorb the entity signals that should point at this client. Even the community footprint is contaminated: all 4 Reddit mentions found are actually about unrelated same-name brands, and 0 mentions about anyone appeared in the last 90 days.
Content citation analysis
What engines cite instead of this domain.
booking.com (cited by all 4 engines), virtuoso.com (3), hotels.com (3), plus travelleaders.com, expedia.com, kayak.com. These are the category authorities AI answers are built from, and this domain is missing from that set entirely.
Holding your pages back
- 0 JSON-LD blocks across the 5 pages checked. No Organization, Article, FAQPage, BreadcrumbList, or WebSite schema. No speakable markup.
- No meta description on the homepage.
- No llms.txt (request returns 404).
- No dateModified or Last-Modified freshness signals.
- No HTML tables or ordered lists, the formats AI engines extract most reliably.
Already working, keep it
- All AI bots are allowed: GPTBot, ChatGPT-User, Claude-Web, PerplexityBot, Google-Extended and others all return HTTP 200. Discoverability scored 27 of 30.
- Clean robots.txt with a sitemap directive and zero syntax errors.
- Server-rendered WordPress, so bots see full content.
- Sound skeleton: single H1, 8 valid H2 sections, 1 question heading, 4 of 8 sections open with a concise answer, 17 external links.
Create the missing identity layer (schema, llms.txt, FAQ page). Update existing pages with meta descriptions, freshness signals, and extractable formats. Amplify by correcting third-party listings so external signals stop feeding the wrong entity.
Key findings
AI engines describe a different company when asked about you.
Evidence. 5 entity-confusion flags, 3 high severity. Perplexity presented two same-named agencies in other countries as likely matches. Gemini led with a same-named LLC registered in 2024 under different founders.
Implication. Any marketing that increases AI queries about the name currently sends some of that attention to other businesses. Disambiguation is the prerequisite for everything else in this report.
Note for our sync
Confirm whether the 2024 LLC is a separate business or your own registration. Site content indicates a founding date roughly 29 years earlier; the answer determines whether this is a factual error to correct or an entity collision to disambiguate.
Zero structured data and no llms.txt, so engines have nothing authoritative to verify you against.
Evidence. 0 JSON-LD blocks on all 5 pages checked. llms.txt returns 404 and scored 0 in all six categories.
Implication. Engines fall back to third-party sources, which is exactly where the same-name confusion comes from. This is also the fastest score recovery available: two tools at 0 with fixes measured in hours.
You are invisible in buying-stage questions.
Evidence. 0 of 8 competitive query mentions. 0 percent share of voice. Rank 119 of 118. Domain absent from every engine’s source set.
Implication. When a traveler asks an AI for agency recommendations, you are not losing the comparison. You are not in it.
What engines do say about you contains material errors.
Evidence. Two inaccuracies flagged. High: a founding year roughly 29 years later than site content indicates. Medium: a destination specialty the company does not actually offer.
Implication. Correctable via the same identity layer (schema founding date, llms.txt corrections), and worth re-measuring after the fix.
Your technical foundation is fine, so the fixes are content-shaped, not infrastructure-shaped.
Evidence. Full bot access, valid robots.txt with sitemap, server-rendered pages, sound heading structure.
Implication. No re-platforming, no developer-heavy project. This is a focused markup and content effort.
Recommended actions
Ordered by impact against effort. Every Radar finding ships with a fix prompt you paste into Claude, ChatGPT, or Cursor.
| # | Action | Impact | Effort | Priority |
|---|---|---|---|---|
| 1 | Organization schema on the homepage: name, url, logo, description, founder, foundingDate, address, contactPoint, sameAs links, plus a disambiguatingDescription stating what you are and are not. | High | Low | Week 1 |
| 2 | Publish llms.txt: H1, blockquote summary, entity definition, product lines, key URLs, and explicit corrections for the founding-date and destination errors. | High | Low | Week 1 |
| 3 | Add meta descriptions site-wide (homepage currently has none). | Medium | Low | Week 1 |
| 4 | FAQ page with FAQPage schema; convert more H2s to question format; keep answer-first openings. | High | Medium | Weeks 2-3 |
| 5 | Add WebSite, BreadcrumbList, and Article schema (with dateModified) and a Last-Modified header; add comparison tables and ordered lists to key pages. | Medium | Medium | Weeks 2-3 |
| 6 | Correct third-party listings (LinkedIn, BBB, Yelp, directory profiles) so external signals match your schema exactly. | High | Medium | Weeks 2-4 |
| 7 | Target long-tail AI prompts in your specialty niches with definitive content rather than competing on "best travel agency" head terms. | Medium | High | Ongoing |
| 8 | Re-establish recent community presence; current footprint shows 0 mentions in 90 days and the existing ones concern other brands. | Medium | High | Ongoing |
Next report and what to watch
A 33 will not move without re-measurement. We recommend a follow-up audit approximately 30 days after the Week 1-3 fixes ship. Watch for:
- 1.Brand Disambiguation climbing off 20 as engines pick up the disambiguating schema. This is the leading indicator.
- 2.Schema Audit and llms.txt moving from 0 immediately after publication.
- 3.Hallucination flags resolving: the founding-date and destination claims should disappear once authoritative signals exist.
- 4.Answer Engine citations beyond Perplexity: ChatGPT and Gemini citing the domain directly is the milestone that says the identity layer landed.
- 5.Share of voice appearing in niche prompts. Expect movement in specialty queries long before generic "best agency" ones.
Run your own
This is what a 33 looks like. What does your domain score?
The free check runs the six technical readiness tools against your domain and previews the rest. Same audit, same scoring, same fix prompts.
