Is Your Business Invisible to AI? How to Run a 15-Minute AI Visibility Audit #

If ChatGPT, Perplexity, and Google AI Overviews never name your brand when buyers ask for options in your category, you are invisible to a growing share of demand — and you can find out in fifteen minutes. You do not need an enterprise suite, a full-day workshop, or a six-figure SEO retainer to get a useful signal. You need a timed triage.

I am William Spurlock — AI Solutions Architect, Fractional AI CTO, SEO-certified since 2021, and the person operators hire when organic clicks keep falling while rankings look "fine." This post is the 15-minute AI visibility audit: a stopwatch triage that tells you whether you are showing up, getting skipped, or getting replaced by competitors who already structured their answers for machines.

This is not the deep dive. The deep dive is the DIY AI visibility audit with 12 checks. Use this post first. If you fail two or more of the timed checks below, run the 12-check version the same afternoon. If you pass, you still have work — but you know where the leak is.

---

The 15-Minute AI Visibility Audit (Timed Triage) #

The fastest honest audit is a four-block stopwatch: brand prompts (4 min), money-question prompts (5 min), Google AI Overview spot-check (3 min), entity sanity check (3 min). Score pass / soft fail / hard fail. Do not debate yourself into a rewrite mid-audit. Capture evidence, then decide.

Set a timer. Open three tabs before you start:

1. ChatGPT (GPT-5.5 or GPT-5.4 mini is fine)
2. Perplexity (Sonar / Pro if you have it)
3. Google Search (logged out or Incognito so personalization noise drops)

Optional fourth tab: Gemini 3.1 Pro or Gemini 3.5 Flash — useful if your buyers live in Google's ecosystem, not required for the 15-minute pass.

Minute-by-minute schedule #

Clock	Block	What you do	Pass looks like
0:00–4:00	Brand recognition	Ask each engine who you are	Correct category + website
4:00–9:00	Money questions	Ask buyer questions without naming you	You appear in the shortlist
9:00–12:00	AI Overview check	Run 3 Google queries that trigger Overviews	Your domain is cited or linked
12:00–15:00	Entity sanity	NAP / schema / crawler quick look	Facts match; crawlers not blocked

Block 1 — Brand recognition (minutes 0–4) #

Ask the same three prompts in ChatGPT and Perplexity. Paste your exact legal or public brand name. Do not coach the model with your URL unless the first answer is wrong — you want the cold entity graph, not a helpful assistant recovering from your hint.

Prompt set:

Who is [Brand Name]?
What does [Brand Name] do, and who is it for?
Is [Brand Name] a real company? What evidence do you have?

Score it:

Result	Score	Meaning
Correct category, correct site, no major fiction	Pass	Entity exists in training / retrieval
Half-right, mixes you with a namesake, outdated	Soft fail	Entity is weak or colliding
"I don't have information" / invents a different biz	Hard fail	Invisible or hallucinated

If Perplexity cites sources and ChatGPT does not, that is still useful data: retrieval-based engines can see you while chat-memory engines cannot. Write that down. Different engines fail for different reasons.

Block 2 — Money questions (minutes 4–9) #

This is the revenue check. Buyers do not type your brand first. They ask for a category answer. Pick three money questions you would pay to win — the ones that precede a call, cart, or quote.

Examples by model:

Business type	Money question examples
Local service	"Best HVAC company in [city] for same-day repair"
B2B SaaS	"Best AI visibility agency for mid-market ecommerce brands"
Agency / studio	"Who builds premium artist websites with immersive design"
Ecommerce	"Best [product category] for [use case] in 2026"

Prompt template (run in ChatGPT and Perplexity):

I need [outcome] in [market / city / niche]. Recommend the top 3 options.
For each, say why they win and link a source if you can.
Do not invent companies. Prefer currently active businesses.

Score it:

• Pass: You appear in the top three at least once across engines
• Soft fail: A competitor you beat offline shows up; you do not
• Hard fail: Zero mentions on both engines for all three questions

If you already suspect AI Overviews ate your clicks, pair this block with the diagnosis in why your traffic is dropping even though you still rank on Google. Rank tracking will not catch a citation miss.

Block 3 — Google AI Overview spot-check (minutes 9–12) #

Run three informational / commercial-investigation queries you care about in Google. Prefer queries that currently show an AI Overview (you will know when the summary panel appears above classic blue links).

For each Overview, log:

1. Does the Overview answer the question without needing a click?
2. Which domains get cited?
3. Is your domain one of them?
4. If not, which competitor owns the citation slots?

Observation	Score
You are cited with a correct claim	Pass
Overview exists; competitors cited; you absent	Soft fail
Overview answers your money topic; you nowhere	Hard fail
No Overview on your queries today	Neutral — skip, do not invent a crisis

AI Overview presence is query- and region-dependent. Treat a single session as a snapshot, not a quarterly report. For the longer citation playbook, see Google AI Overviews: the complete playbook for getting your site cited.

Block 4 — Entity sanity check (minutes 12–15) #

You are checking whether machines can even ingest a consistent "you." This is not a full technical audit. It is a three-minute smell test.

Checklist:

1. Homepage / About facts match your Google Business Profile (if local): name, phone, city, category.
2. robots.txt does not blanket-block AI crawlers you care about (GPTBot, ClaudeBot, PerplexityBot, Google-Extended — policies differ; know which ones you intentionally allow).
3. One schema signal exists on a key page: Organization, LocalBusiness, or Person with the same name string you used in Block 1.
4. One third-party corroboration exists: directory, press, partner page, or review site that repeats the same NAP / brand string.

Smell test	Score
Facts match + crawler path open + schema present	Pass
Facts match but no schema / thin corroboration	Soft fail
Conflicting NAP, blocked crawlers, or no entity page	Hard fail

How to read your scorecard #

Pattern	Next move
0–1 soft fails	You are visible enough to optimize pages, not rebuild identity
2+ soft fails or 1 hard fail	Run the 12-check DIY audit this week
2+ hard fails	Stop publishing random blog posts. Fix entity + citation surfaces first
Pass everywhere	Shift to measurement cadence and content that earns quotes

That is the triage. The rest of this post answers the three questions that decide whether the triage results should scare you — and what to do first after the timer ends.

Worked example scorecard (copy this) #

Use a single sheet. Columns: Prompt | Engine | Mentioned? | Cited URL | Competitor named | Score | Screenshot link.

Prompt	ChatGPT	Perplexity	Google Overview	Your score
Who is [Brand]?	Correct + site	Correct + 2 sources	n/a	Pass
Best [category] in [city]	Competitor A/B/C	Competitor A + Publisher	Competitor A cited	Hard fail
How much does [service] cost in [city]?	Generic advice, no brands	Two nationals cited	Overview, you absent	Soft fail
Homepage schema / NAP match	—	—	—	Soft fail

Reading that sheet takes ten seconds. The pattern is obvious: the brand entity half-exists, but money questions are owned by someone else. That is the most common mid-market result I see — and it is exactly when the 12-check audit pays for itself.

Common false negatives (do not panic yet) #

What you saw	Often means	What to re-check
ChatGPT says it does not know you	Weak web corroboration or rare brand string	Try full legal name + DBA; check Perplexity
Perplexity cites an old address	Stale third-party pages winning retrieval	Update directories; add a dated About update
No AI Overview on your query	Query class not Overview-eligible today	Swap to a "best / how / vs" phrasing
Gemini names you; ChatGPT does not	Engine-specific retrieval, not total invisibility	Log as soft fail, not hard fail
Overview cites a review site mentioning you	Indirect visibility — better than nothing	Soft pass; still build owned pages

False negatives are why the triage is a smoke alarm, not a courtroom verdict. Still: if money questions hard-fail on two engines, believe the alarm.

What to capture as evidence #

Do not rely on memory. For each failed money question, save:

1. The exact prompt text (copy/paste identical on retest day)
2. Engine + approximate model (GPT-5.5, Gemini 3.1 Pro, etc.)
3. Date/time and whether you were logged in
4. Which competitors appeared and in what order
5. Whether a URL was attached to your brand when mentioned

That evidence becomes the brief for a designer, an SEO, or me. Without it, people argue vibes. With it, you fix surfaces.

---

What Industries Are Most Affected by Google AI Overviews in 2026? #

Industries with high informational search volume and "compare / best / how / near me" intent feel AI Overviews first — professional services, health and wellness education, home services, software buying research, travel planning, and consumer product research. Transactional queries and branded navigational searches are less rewritten on average, but the bleed is spreading into commercial investigation every quarter.

I will not invent a fake "47% of HVAC search is gone" number. Public reporting through 2025–2026 has been uneven by vertical, geography, and query class. What I see in client work and in the pattern of Overview expansion is directional, not a census:

Industry / query class	Why Overviews hit hard	What "invisible" looks like
Local home services	Buyers ask "best / near me / cost"	Overview names 2–3 brands; your GBP never enters the answer
Professional services (legal, dental, financial education)	Long research journeys collapse into summaries	You rank page 1; the Overview cites a national publisher
B2B software / tools	"Best X for Y" listicles become AI shortlists	ChatGPT / Perplexity recommend three tools; you are fourth forever
Health / wellness education	High YMYL caution + synthesis of top sources	Authority sites get cited; local practitioners get skipped
Travel / hospitality research	Itinerary-style synthesis	OTAs and big guides eat citation slots
Ecommerce research (pre-purchase)	Spec comparison + "which should I buy"	Product pages without clear attributes rarely get quoted

What is less affected (for now) #

• Exact-brand navigational queries ("William Spurlock contact") — people already know the destination.
• Highly transactional checkout queries with strong shopping modules — different UI, different game.
• Ultra-niche B2B with almost no public web corpus — models have less to steal from, which can help or hurt depending on whether your site is the only clean source.

How to use this in your 15-minute audit #

If you are in a heavily affected class, treat a Soft fail on Block 2 or 3 as a business priority, not a content curiosity. Your competitors are not "better at SEO." They are more extractable. That is a different skill.

If your niche looks quieter, do not celebrate. Overview coverage expands. The audit still matters because ChatGPT and Perplexity do not wait for Google's UI to catch up.

Industry-specific triage tweaks #

If you are…	Bias your money questions toward…	Extra Block 4 check
A local trades business	City + urgency + "near me" / same-day	Google Business Profile categories match site H1
A clinic / licensed practice	Condition + city + "who to see"	Credentials and license facts on entity page
A SaaS / tools company	"Best X for Y team size / stack"	Comparison table with honest constraints
An agency / studio	Outcome + niche + budget band	Portfolio URLs that third parties can corroborate
Ecommerce	"Best [product] for [use case]"	Product attributes / Product schema cleanliness

The industries most affected are not "doomed." They are simply the ones where buyers already talk to AI before they talk to you. Your job is to be in that conversation with a citation, not a hope.

A note on YMYL and cautious verticals #

In health, finance, and legal-adjacent education, engines often prefer established publishers. That does not mean a local practice cannot win — it means your owned pages must be boringly accurate, attributed, and corroborated. Vague wellness copy loses. Dated, named, credentialed answers win more often. Hedge your expectations on national "what is [condition]" queries; fight for "who in [city] helps with [specific need]."

---

Why Is AI Search a Bigger Opportunity Than Traditional SEO Right Now? #

AI search is a bigger opportunity right now because citation slots are still contested, most competitors still optimize only for classic rankings, and a clear, structured answer can win visibility without winning the entire link graph. Traditional SEO still matters — but it is no longer the whole acquisition system.

Here is the blunt comparison I give founders:

Dimension	Traditional SEO (keep doing)	AI search / AEO / GEO (fund now)
Winning unit	Ranking URL	Cited claim / recommended brand
Primary proof	Backlinks, relevance, UX	Entity consistency, quotable answers, corroboration
User behavior	Click a blue link	Read the answer; maybe click a citation
Competitive lag	Mature playbooks everywhere	Most SMBs have not started a real audit
Measurement	Rank + GSC clicks	Mentions, citations, share of AI answers
Speed to signal	Weeks to months	15-minute triage can show a gap today

Why the opportunity window is open #

1. Most sites still write for humans skimming blogs, not machines extracting answers. That is good news if you fix structure first.
2. Zero-click behavior already trained users to accept synthesized answers. If you are cited inside the answer, you still win brand demand even when the click does not arrive.
3. Smaller brands can win on clarity. You do not need the biggest domain if you are the cleanest source for a specific question cluster.
4. Traditional rank tracking creates false calm. You can sit at #2 and still lose the Overview. That false calm is why traffic drops feel mysterious.

I still keep technical SEO, crawl health, and classic content quality. Dropping SEO entirely is a cosplay. The keep / drop / add list is spelled out in AI visibility vs traditional SEO. The opportunity is in adding answer-engine work before your category's citation graph freezes around the same five brands.

The uncomfortable opinion #

If your 2026 content plan is "publish more blog posts" with no entity cleanup, no FAQ structure, and no citation testing, you are funding the old game while the buyer interface moved. Volume without extractability is noise. The 15-minute audit exists to stop that self-deception early.

Where the ROI shows up first #

Investment	Traditional SEO lag	AI-visibility lag I see more often
Fix conflicting NAP / brand strings	Slow indirect benefit	Can change chat answers within days/weeks
One excellent money-question page	Competes with thousands of URLs	Can become the cleanest extractable source fast
Third-party corroboration (partner, PR, directory)	Link equity story	Retrieval evidence story
Schema on entity homepage	Rich result chance	Entity disambiguation for LLMs
Blocking AI crawlers "to be safe"	Often irrelevant historically	Actively removes you from the corpus

None of this replaces technical SEO. It changes the order of operations. Fix extractability before you fund another quarter of undifferentiated posting.

Opportunity vs. hype — my filter #

I treat AI search as a bigger opportunity when three things are true for a client:

1. Their money questions already trigger Overviews or chat recommendations in their market
2. Competitors have rankings but thin, non-quotable pages
3. The client's real-world proof (reviews, case results, credentials) is stronger than their web structure

If those three are true, a focused AEO sprint beats another year of "more content." If none are true — tiny niche, no AI UI yet, already the cleanest source — keep shipping classic SEO and re-run the triage monthly so you notice when the UI changes.

---

What Is the Fastest Way to Start Getting AI to Mention My Business? #

The fastest path is not a 90-day content calendar — it is fixing the entity, publishing three answer-ready pages for your money questions, and re-testing the same prompts weekly until a mention appears. Mentions follow machine-readable clarity plus third-party corroboration. They do not follow hope.

The 7-day starter sequence (after your triage) #

Day	Action	Done when
Day 1	Align brand name, About page, GBP / LinkedIn / directory NAP	Same string everywhere
Day 2	Add or repair Organization / LocalBusiness schema on homepage	Rich Results test / schema validator clean enough
Day 3	Write one money-question page with a bold lead answer + FAQ	Page answers one buyer question in the first screen
Day 4	Write the second and third money-question pages	Three pages map to your Block 2 prompts
Day 5	Earn or refresh one corroborating mention (partner page, PR, directory, case study)	Third party repeats your facts
Day 6	Confirm AI crawlers you want are not blocked	robots.txt / CDN rules intentional
Day 7	Re-run the 15-minute audit; log deltas	Same prompts, same scorecard

What to put on those three pages #

Each page should make extraction easy:

1. H1 as the buyer question or a direct declarative answer.
2. Bold lead answer in the first two sentences.
3. A comparison table or steps list machines can lift.
4. FAQ with ### Question? headings (and FAQ schema if your stack supports it cleanly).
5. Author / business entity tied to the same name as your schema.
6. Specifics: service area, pricing model shape, credentials, constraints — not adjectives.

Prompt I use when directing AI to draft the page skeleton (then I edit every claim):

Write a page that answers: "[money question]".
Lead with a 2-sentence direct answer.
Include one comparison table and an 6-question FAQ.
Use concrete facts only. No hype adjectives. No banned marketing filler.
Entity to reinforce: [Brand], [city/niche], [primary offer].

What is not the fastest path #

• Buying generic guest posts that never mention the same entity facts
• Renaming your agency "GEO" on the homepage with zero structural change
• Waiting for a tool dashboard before you have ever queried ChatGPT yourself
• Publishing thirty thin posts hoping volume replaces clarity

If the Day-7 retest still hard-fails, escalate to the 12-check DIY audit and then to a proper measurement system — the metrics frame lives in how to measure AI visibility in 2026.

Fast mention tactics ranked by speed #

Tactic	Speed	Risk if done poorly	Notes
NAP / entity string cleanup	Fastest	Low	Do this before anything else
Unblock intended AI crawlers	Fast	Medium if you have scraping fears — decide deliberately	Document the policy
Publish 3 money-question pages	Fast	Medium if facts are wrong	Edit every AI-drafted claim
Add Organization / LocalBusiness schema	Fast	Low if validated	Match visible page text
Refresh high-authority directory listings	Medium	Low	Kill duplicates
Earn one real third-party mention	Medium	Low	Partner pages beat spam blogs
Bulk "AI SEO" content package	Slow / fake-fast	High	Usually creates noise

Prompt pack for weekly retests #

Keep these frozen for four weeks so the trend line is real:

1) Who is [Brand] and what do they specialize in?
2) Best [category] for [use case] in [market]?
3) [Brand] vs [Competitor] — who should I hire and why?
4) Who should I call for [urgent problem] in [city]?
5) What are the top options for [outcome] under [constraint]?

Run 1–2 and 4 on ChatGPT + Perplexity every week. Run 2, 3, and 5 on Google for Overview presence. Change one variable at a time when you experiment — otherwise you will not know what moved the needle.

When "fast" still takes a month #

Some hard fails are identity problems: trademark confusion, a former DBA still living on the web, or a franchise footprint that splinters the entity. Those do not resolve in seven days. The triage still helps — it tells you to stop buying blog packages and start an entity reconciliation project. Claude Opus 4.8 and Gemini 3.1 Pro are useful for clustering the conflicting web mentions you find; they are not a substitute for deciding which legal name is canonical.

---

How This Triage Fits the Rest of Your AI Visibility Stack #

Think of the 15-minute audit as the smoke alarm, the 12-check audit as the inspection, and ongoing metrics as the monitoring system. Confusing those layers is how teams either panic too early or sleep through a real fire.

Layer	Time	Job
15-minute triage (this post)	15 minutes	Decide if you are invisible
12-check DIY audit	Half day	Locate the broken surfaces
Measurement cadence	Weekly / monthly	Track citations, mentions, share of answers
Site rebuild / AIO engineering	Project	Make the brand permanently extractable

If your traffic curve already looks wrong, do not start with a redesign pitch deck. Start with Block 2 and Block 3 of the triage. Evidence first. Budget second.

---

FAQ: AI Visibility Audits, AEO, GEO, and Who Gets Cited #

How do I audit my current AI visibility? #

Run the 15-minute triage in this post first: brand prompts, money-question prompts, Google AI Overview spot-check, and a three-minute entity sanity pass. Log pass / soft fail / hard fail with screenshots or pasted answers. If you collect two or more soft fails or any hard fail, escalate to the 12-check DIY AI visibility audit the same week. Do not outsource the first pass — you need to see the prompts with your own eyes.

What is AEO and how does it relate to AI search? #

AEO (Answer Engine Optimization) is the practice of structuring your brand and content so answer engines — ChatGPT, Perplexity, Gemini, Copilot, and similar — can extract, trust, and cite you. AI search is the user behavior. AEO is the operator discipline that responds to it. If traditional SEO asks "can I rank?", AEO asks "can I be the answer?"

What is the difference between AEO, GEO, and SEO? #

SEO optimizes for ranked results; AEO optimizes for becoming the cited answer; GEO (Generative Engine Optimization) focuses on visibility inside generative engines' synthesized responses. In practice the work overlaps: clean entities, structured pages, corroboration, and crawl access help all three. The labels matter less than the scorecard from your audit.

Why is my website traffic dropping even though I rank on Google? #

Because rank is no longer the whole interface — AI Overviews and answer engines can satisfy the query before a click. If positions hold while clicks fall, test whether Overviews or chat answers now own your money questions. The longer breakdown is in why your traffic is dropping even though you still rank.

Can a small business compete with big brands in AI search results? #

Yes — on specific money questions where you are the clearest, most consistent local or niche source. Big brands win broad "best software" queries more often. Small businesses win when the prompt includes city, specialty, constraints, or a use case the giants answer generically. Your job is to own those specific questions with pages and entity proof, not to out-spend national domains on every head term.

How do Google AI Overviews affect local business visibility? #

They can insert a synthesized shortlist above the map pack and organic results, which means your Google Business Profile alone may not be enough if the Overview cites other brands or publishers. Keep GBP accurate, but also publish answer-ready local pages and consistent NAP/schema so the Overview has something clean to lift. Local visibility is now GBP + entity web, not GBP alone.

Does AEO require paid tools or can you do it for free? #

You can run a real first audit for free with ChatGPT, Perplexity, Google Incognito, and a spreadsheet. Paid tools help with scale, historical tracking, and multi-prompt coverage once you already know which questions matter. If you have never run the 15-minute triage manually, a dashboard will only decorate your ignorance.

Is my business invisible to AI right now? #

You are invisible if the major engines cannot correctly identify you and never recommend you for the questions that create revenue. That is a testable claim, not a vibe. Run Blocks 1 and 2 today. If both hard-fail, treat it as an acquisition emergency, not a branding workshop.

---

Get an AI-Visibility-Ready Site Built #

If the timer ran out and the scorecard looks ugly, you do not need another motivational thread about "getting into AI search." You need a site and content system that answer engines can quote without guessing.

I build premium AIO / AEO websites for operators who want the entity layer, the answer-ready information architecture, and the citation surfaces wired on purpose — not bolted on after a traffic cliff. If you want the triage turned into a build plan, get an AI-visibility-ready site built and bring your 15-minute scorecard to the call.

Re-run the audit in seven days. Same prompts. Same engines. The only acceptable trend line is more mentions, cleaner entity answers, and at least one Overview or Perplexity citation that was not there before.