AI's local answer surface is 90% metro-specific — and the metro's own media is almost none of it

What we measured, and how

The corpus is 173 completed audits — 1,024 answers (med spas, personal-injury law firms, dentists, home services; 42 US metros; audits run 2026-06-12 to 2026-07-14). Synthetic QA rows are excluded by an exact-name filter, not a regex over the vertical field. For this cut we took every completed answer, extracted every raw URL from the answer text, collapsed each to its registrable domain, and tagged that domain with the metro of the business the question was about. The counting unit is answers-citing-a-domain, not links — an answer counts once for a domain no matter how many times it links it.

One limit belongs at the top: this is ChatGPT only. All 511 of our ChatGPT answers carried at least one link. Our 513 Perplexity answers carried 13 links between them, because we capture the model's text output through the API rather than the citation panel its interface renders. That is an instrument limit of ours and says nothing about what Perplexity reads.

Finding 1: 980 of 1,092 cited domains appear in exactly one metro

89.7% of the citation surface is metro-exclusive. Only 112 domains (10.3%) were cited in more than one city, and those are the names an SEO would already guess — reputation platforms and national aggregators. The table below is every domain cited in 13 or more of our metros.

Of the 980 metro-exclusive domains, 95 are the audited business's own website. The other 885 are third-party sites that surfaced in one city and never again — competitors' sites, single-city directories, local blogs and forum threads. That tail is where most of a local answer's sources come from, and almost none of it is reusable from one city to the next.

Finding 2: the city's own press is 6 citations out of 379 answers

If the surface is local, city media is the intuitive candidate — a "best med spas in Nashville" feature is exactly the shape of page an engine ought to reach for. We took our 15 largest metro cuts and, for each, checked the leading city magazine, the metro daily and the alt-weekly: 46 domains in total, listed in full below so the check is repeatable. Against the 379 ChatGPT answers from those metros they were cited 6 times.

The 46 domains checked, three per metro: charlottemagazine.com, charlotteobserver.com, qcnerve.com · nashvillelifestyles.com, tennessean.com, nashvillescene.com · tampamagazines.com, tampabay.com, cltampa.com · 5280.com, denverlifemagazine.com, denverpost.com, westword.com · austinmonthly.com, statesman.com, austinchronicle.com · dmagazine.com, dallasnews.com, dallasobserver.com · atlantamagazine.com, ajc.com, creativeloafing.com · vegasmagazine.com, reviewjournal.com, lasvegasweekly.com · houstoniamag.com, chron.com, houstonpress.com · sandiegomagazine.com, sandiegouniontribune.com, sdreader.com · phoenixmag.com, azcentral.com, phoenixnewtimes.com · raleighmag.com, newsobserver.com, indyweek.com · kansascitymag.com, kansascity.com, thepitchkc.com · orlandomagazine.com, orlandosentinel.com, orlandoweekly.com · columbusmonthly.com, dispatch.com, columbusalive.com.

What this does not prove

• It is not a ranking of publications. A city magazine can be authoritative, widely read and completely absent from this corpus. Absence here means ChatGPT did not attach a link to it in these particular answers — nothing more.
• Our list of 46 is a judgment call. We picked three obvious outlets per metro. A different three could move the count. The list is printed above precisely so anyone can rerun it against their own.
• Four verticals only. Med spas, injury law, dentists and home services. "Best restaurants in Austin" is exactly the query where city media should dominate, and it is not in our corpus.
• Small per-metro n. Orlando is 6 answers and Kansas City 9. A zero at that sample size is weak evidence on its own; the strength of the finding is the 15-metro total, not any single row.
• ChatGPT only, one month, one instrument. See the limit at the top.

Why this matters to anyone selling local visibility

• A citation-surface audit does not port between cities. 89.7% of cited domains appeared in one metro only. The Charlotte source list is not the Dallas source list, in the same vertical, for the same question.
• "Get press coverage" is not, on this evidence, an AI-visibility tactic in these categories. Six citations across 379 answers is a rounding error next to birdeye.com's 134.
• The reusable part of the surface is 112 domains wide. That is the portion of the work an agency can standardise across a book of clients; everything else is per-city research.

The takeaway

The engines answer local questions with local sources — just not the local sources a city would name. Nine of every ten domains they cite show up in one metro and never again, and the metro's own magazine, paper and alt-weekly account for six citations across 379 answers.

The free 60-second check runs a business's own buyer questions and shows what the engines cite when they answer. Agencies mapping a citation surface across several cities can run five businesses at once with the $249 Agency 5-pack. If you publish one of the 46 domains above — or any of the 1,092 we logged — and want the exact answers that did or did not cite you, ask us and we will send them.