Will ChatGPT Recommend Your Company? A 15-Minute Test Any Owner Can Run

You heard it on a lost job: “ChatGPT said to call the other guy.”

Maybe you’re in the Map Pack some weeks. Maybe Tuesday went quiet and you don’t know if AI even knows your company exists.

Stop guessing. In about 15 minutes you can run the same non-branded prompts homeowners use, screenshot what comes back, and see whether you’re named, linked, ignored — or only showing up for the wrong service.

This is a hands-on owner ritual. Not a case-study flex. Not a promise ChatGPT will name you next month. Just a clear test, a way to read results, and a 30-day fix list if you’re invisible.

Why a 15-minute test beats guessing

“Someone said AI recommended a competitor” is a rumor until you reproduce the prompt.

BrightLocal’s Local AI Visibility Study (Sep 16, 2026) showed why guessing fails: Google Maps mentioned tracked businesses in about 66% of searches, while AI surfaces (AI Overview, ChatGPT, AI Mode) only about 32–38%. Same “who should I call” intent. Different shortlists. ChatGPT overlapped only about 19–20% with Google’s AI surfaces on named businesses.

Your Map Pack win does not travel automatically. We unpacked that gap in You Can Rank in Google’s Map Pack and Still Lose the Job to ChatGPT. Today you run the DIY check.

Homeowners aren’t ditching Google. BrightLocal’s Consumer Search Behavior — Channels study (Jul 9, 2026) found 75% used more than one channel on their last local search; AI showed in 23% of those journeys; among AI users, only 18% felt ready to contact without verifying elsewhere. AI shortlists. Maps and reviews still close a lot of deals.

Soft context from BrightLocal’s Local Consumer Review Survey 2026: 45% used generative AI for local recommendations in the past year (up from 6%). That’s why the line is getting louder — not why you abandon Maps.

For SEO / AEO / GEO labels without the fog: SEO vs AEO vs GEO for local businesses. This piece stays on the test.

Exact prompts to run (trade + city; variants; what not to ask)

Open ChatGPT (note logged-in vs guest). Use non-branded prompts — trade + city, the way a stranger asks. Fill the brackets with your real trade and city.

Core set (copy-paste)

1. Who are the best [HVAC / plumbing / electrical] companies in [City, ST]?
2. I need an emergency [AC repair / plumber / electrician] in [City] — who should I call?
3. Recommend a reliable [water heater install / panel upgrade / furnace repair] company near [City or neighborhood].
4. What [HVAC / plumbing / electrical] companies in [City] have strong recent reviews?

Variants

5. Best [trade] near me for [City] homeowners — short list with why.
6. Compare top [trade] options in [City] for same-day service.

Run three to four prompts that match your money jobs. HVAC: AC repair / furnace / install. Plumbing: emergency / water heater / drain. Electrical: panel / EV charger / outage.

What not to ask

Don’t ask: Is [Your Exact Legal Name] the best company in [City]? — branded ego check. Homeowners don’t start there.

Don’t paste your website and beg for a recommendation. Spamming AI with self-promotional prompts is not a growth hack — and it won’t show what a stranger sees.

Don’t judge from one odd phrasing. Variance is real (next section).

Optional later: one pass in Perplexity or Gemini. Start with ChatGPT so the first pass is clean and comparable.

How to screenshot and what to record (named? linked? competitor only?)

Screenshot each full answer (including source links or map cards). Log a simple row:

Score honestly:

• Named — your company or clear DBA appears.
• Linked — a URL or GBP-style card points to you (not a generic directory dump).
• Competitor only — others get the shortlist; you’re absent.
• Wrong service — named for work or a city you don’t cover.

Note hallucination flags: wrong phone, closed location, services you don’t offer. Keep the screenshots for the free Visibility Report.

Run it twice: luck vs persistence (BrightLocal variance / persistence)

One answer is a coin flip. Two passes tell you more.

BrightLocal’s Local AI Visibility Study (Sep 16, 2026) documented high variance across prompts and surfaces — who shows up is not a locked leaderboard. Persistence is the counterweight: when a business did appear once, it reappeared about half the time or more on repeat prompts (~50%+). Volatility is real; total chaos is not.

Owner ritual:

1. Run three core prompts today. Screenshot. Log the table.
2. Wait 24–48 hours (or a fresh chat). Run the same three wording again.
3. Compare: named both times, once, never — and whether you were linked.

• Named both → you’re in the pool for that query shape (still watch competitors).
• Named once → luck or weak persistence; fix evidence, retest in 30 days.
• Named never → treat as invisible for that prompt family until the 30-day work lands.

Do not hammer the model with branded “recommend me” prompts hoping to train it. Measure stranger discovery.

How to interpret results (you’re missing / you’re inconsistent / you’re cited for the wrong service)

You’re missing

Nobody names you. Usual causes: thin or missing service pages, weak review authority, almost no third-party mentions, or a Google Business Profile leaking basics (wrong category, vague services, stale hours). AI has little clean material to cite.

BrightLocal’s citation mix: business websites ≈ 42% of AI citations; GBP ≈ 28.63%. A “we do it all” homepage with no dedicated money-service pages starves both.

You’re inconsistent

You flicker across the two runs. That matches BrightLocal’s variance + ~50%+ persistence pattern. Strengthen the same stack — service pages, recent specific reviews, real mentions — so repeats have something stable to grab.

You’re cited for the wrong service

You’re named for work you don’t sell or a city you don’t cover. Fix labels first: categories, services list, page titles, third-party blurbs.

If Maps looks fine some weeks while AI ignores you, that’s the Maps≠AI gap — not a reason to abandon the pack. If the phone got quieter even when you “still rank,” that’s a different leak (calls vs pin) — see Why Your Phones Got Quieter Even Though You Still “Rank”.

GBP foundation either way: Google Business Profile audit — 12 silent leaks.

The 30-day fix list if you’re invisible (service pages, reviews, mentions — ordered)

Same order as the Maps≠AI close-the-gap plan. Full narrative: Rank in the Map Pack and still lose to ChatGPT.

Whitespark’s Official 2026 Local Search Ranking Factors scored AI search visibility separately. Experts leaned toward expert best-of lists, dedicated service pages, industry-domain prominence, quality unstructured citations, and review authority — not GBP posts alone.

Days 1–10: Service pages

One clear page per money service: what you fix, who it’s for, cities/neighborhoods, booking path, proof, short FAQs (“Same-day?”, “Repair or replace?”). Soft deeper dive: Generative Engine Optimization and Answer Engine Optimization.

Days 11–20: Review authority

Ask for recent, specific Google reviews and at least one strong third-party shelf. Coach: name the job and town. Reply like a human. Rating floor + last-90-day freshness matter for trust — see 4.5 Stars and Reviews From the Last 90 Days: The New Local Trust Bar.

Days 21–30: Mentions (quality over spam)

Earn a few real unstructured mentions: association, supplier spotlight, local news, neighborhood blog. Pitch one best-of list where you belong. Fix NAP on directories that already cite you. Skip vanity citation dumps.

Re-run the same three prompts (twice). Update the table. That’s month two’s gap list. Broader help after the free grade: Local SEO services.

CTA — send screenshots / get the Free Local Visibility Report (homepage)

You have prompts, a screenshot checklist, and a way to tell luck from persistence.

Run the test, screenshot the answers, and get a free read on what’s missing — Maps + AI.

Paste what you saw (named / linked / competitor-only / wrong service) when you request the audit on the homepage. We’ll turn the ritual into a fix list — without pretending AI spam is marketing.

Sources

All figures accessed 2026-09-23.

1. BrightLocal — Local AI Visibility Study (Sep 16, 2026). Maps ~66% vs AI surfaces ~32–38%; ChatGPT overlap ~19–20% with Google AI surfaces; high variance; ~50%+ persistence when a business appears on repeat prompts; GBP ≈ 28.63% of citations; business websites ≈ 42%.
   https://www.brightlocal.com/research/local-ai-visibility-study/
2. BrightLocal — Consumer Search Behavior — Channels (Jul 9, 2026). 75% used more than one channel; AI in 23% of most recent searches; only 18% of AI users felt ready to contact without verifying.
   https://www.brightlocal.com/research/consumer-search-behavior-channels/
3. Whitespark — Official 2026 Local Search Ranking Factors. AI visibility weighting: expert best-of lists, dedicated service pages, industry-domain prominence, unstructured citations, review authority (not GBP posts alone).
   https://whitespark.ca/local-search-ranking-factors/
4. BrightLocal — Local Consumer Review Survey 2026 (Feb 11, 2026). Soft context: 45% used generative AI for local recommendations (up from 6%).
   https://www.brightlocal.com/research/local-consumer-review-survey/