A Salt Marsh Growth AI demonstration  ·  See what we build for service businesses at saltmarshgrowth.ai
Salt Marsh Growth AI · AI Audit · Recorded AI answers

When Buyers Ask the AI, Who Gets Named?

Dated public-source observations. Not a client engagement.
We asked the AI engines real buyer questions about Beaufort real estate — for real, in July 2026 — and recorded exactly who they named. This page shows the actual results, and what a firm can do about them.
A working sample of our
AI Audit
All results real · engine & date labeled

Why this matters now

58%
Google's share of combined search + AI discovery — down from 89% in 3 yrs
FirstPageSage 2026
25%+
of Google searches now show an AI Overview — cutting top-result clicks ~58%
Memeburn · Contently
14.2%
conversion rate of ChatGPT referral traffic vs ~2.8% from organic search
Stackmatix 2026
900M
weekly ChatGPT users — buyers now start research in AI conversations
Feb 2026

The panel — five real questions, real answers

A locally-tuned prompt panel, run live. Exact answers vary by session and user — that's why we sample and trend rather than promise. Here is who the engines actually named:

Buyer questionEngineWho was named / cited
"Who are the best real estate agents in Beaufort, SC?"PerplexitySix individual agents named + The Faris Group ("400+ homes, $300M+, 5★"). Sources: US News agent rankings, agent sites, Zillow.
"What is the best real estate brokerage in Beaufort SC?"PerplexityThe Faris Group, Chambers Helms Group (KW), Beaufort Realty Consultants, Collins Group Realty.
"I want to buy a house on Fripp Island — what should I know?"PerplexityAnswer built from two firms' island guides — the firms that wrote the content got the citations. No agent recommendation given.
"What are the short-term rental rules in Beaufort SC?"PerplexityNo local firm cited at all — the authoritative answer is unclaimed.
"Best real estate agents in Beaufort South Carolina"GoogleLocal pack: The Faris Group 5.0★(73), a long-established Lady's Island firm 4.9★(46), BHHS Bay Street 5.0★(82). (No AI Overview rendered in our sample.)

World 1: Google's local pack

Still driven by Google Business Profile + reviews. Beaufort's established firms show up here — though we found real profile errors even among them (one long-established firm's profile says "10+ years in business").

World 2: AI answers

Driven by content, citations, and rankings — a different winner's list. Several of Beaufort's longest-established names never appeared in any AI answer in our panel. Strong on Google ≠ visible in AI.

What we recorded locally:

One Beaufort team was named in three of the five answers we recorded. Why? They publish citable, statistic-rich local guides and run a heavy review cadence. The engines quote what's quotable. What they did by instinct can be done deliberately, once you know what the answers cite.

The open territory:

Topic questions — STR rules, island buying guides, insurance costs — are answered from whoever wrote the best page. In our panel, the STR-rules answer cited no local firm at all. The firm that publishes the authoritative Beaufort answers becomes the firm the AI quotes to every future buyer who asks.

The readiness audit — the five checks we run

The technical scorecard behind every audit. Each check is pass/fail on your actual site — no guesswork, and most firms are surprised by at least one result:

CheckWhy it matters
Reviews, profiles & local trust (GBP + Bing Places)ChatGPT and Perplexity can't see your Google Business Profile — they anchor "who's good here" from indexed reviews, directories, and rankings. Bing matters again (ChatGPT leans on its index).
Citable local contentStatistics +34%, quotations +44% citation lift (Princeton GEO study). Guides get quoted; interchangeable content gets summarized.
AI crawlers allowed (robots / CDN)Engines can't quote a site they can't reach. (Cloudflare users: AI bots now blocked by default.)
Schema markup (entity data)Machine-readable identity — who you are, where, what you do, your reviews.
llms.txt & AI hygieneA curated "start here" file for language models. Google says its AI doesn't need it — we treat it as free hygiene for the other engines, never a headline fix.

We ran these same checks on our own site first and fixed what we found. Every client audit reports exactly this honestly — pass and fail alike.

What we actually do about it

  1. Access: verify every AI crawler can reach the site (robots.txt, CDN settings) — the silent killer we check first.
  2. Entity: schema markup, llms.txt, clean structure — make the firm machine-readable.
  3. Local trust: Google Business Profile + Bing Places completeness, review cadence with owner responses, consistent listings everywhere.
  4. Citable content: the hyperlocal guides buyers actually ask about — written the way engines quote (stats, sources, plain answers).
  5. Measurement: the prompt panel re-run quarterly — your share of AI answers vs named competitors, trended over time, with receipts.
Method & honesty: panel run 23 Jul 2026 on anonymous sessions (Perplexity; Google Search US). AI answers vary by user, session, and day — results are a labeled sample from one date, not a census; a second run would show what moved. We record what the services say; no one controls what they say, so no one can promise you a place in their answers. Named results above are what the engines actually said; firms are named only where the engines named them publicly.