Field Reports

What AI assistants actually say about real practices.

These are real audits of real businesses — what ChatGPT, Claude, Gemini, Perplexity, Google AI Overviews, Grok, and Copilot said when we tested them, what turned out to be broken, and the exact fix plan. Practice details are anonymized. Findings are not.

Field Report · Dental · Central Texas

The two-location dental group whose AI answers pointed patients to the wrong office

July 2026 · anonymized

A dental group with two offices about twenty minutes apart. Solid practice, real reviews, two active locations — and AI assistants couldn't tell them apart.

Finding

When we tested standardized patient prompts ("best dentist near me," "emergency dentist in [city]"), the AI platforms had blended the two offices into one confused identity — anchored on the secondary location's address.

Finding

A direct competitor operating near the second office never surfaced in the practice's competitive picture at all — the confusion made the practice structurally blind to who was winning recommendations in half its market.

Root cause

The website shipped multiple blocks of structured data (schema), and the first one AI systems encounter carried the business name with no address at all. The blocks that did carry full addresses were buried behind it. Machines read in order; patients got the blend.

The prescription

✓One complete LocalBusiness schema block per office — name, address, phone, geo — with the primary location unambiguous.
✓NAP (name / address / phone) consensus cleanup across the directories AI models actually cite for dental queries.
✓Separate, fully-built Google Business Profiles per location instead of one profile stretched across two.
✓Re-test against the same locked patient-prompt set until both locations resolve cleanly.

Why it matters for your practice: if you have more than one location — or ever moved — the entity AI assistants believe you are may not be the one your patients need to find. This failure is invisible from inside your own Google dashboard.

Field Report · Dental · Connecticut

The established practice with a real Google presence — and a 36/100 AI visibility score

July 2026 · anonymized

A long-established dental practice with a legitimate traditional footprint: indexed site, real reviews, years of history. Its audited AI visibility score: 36 out of 100.

Finding

On standardized patient prompts, the recommendation slots went overwhelmingly to national directories — Healthgrades, WebMD-class sites — not to any local practice. The practice wasn't losing to the dentist down the street; it was losing to aggregators.

Finding

The practice's structured data announced its name to machines but not its address — the one block AI systems hit first was incomplete, so the entity couldn't be grounded to a place. An AI that can't place you can't recommend you for "near me."

Root cause

Nothing about the practice was fake or thin — its machine-readable identity was. Traditional SEO signals (rankings, backlinks) don't transfer to AI recommendation engines without entity data those engines can parse.

The prescription

✓Complete Dentist/LocalBusiness schema — full name, street address, phone, hours — on every key page, first in document order.
✓Citation consistency pass across the 25+ directories AI models cite for healthcare queries, so every source agrees on who and where the practice is.
✓Content targeted at the exact prompts where the audit found the practice dark — procedure + city pages matching how patients actually phrase AI questions.
✓Monthly re-scoring against the same locked prompt set to measure movement platform by platform.

Why it matters for your practice:page one of Google and invisible to ChatGPT is the normal state for established practices right now — the signals that earned your Google position are not the signals AI assistants read. The gap is fixable, and it's mostly plumbing.

Full before-and-after client trajectories publish here as Special Ops engagements complete their first 90 days — with real month-over-month scores from the same locked prompt sets, not hand-picked screenshots. We don't publish invented results, so this section grows slowly.

About Special Ops (done-for-you)

Your own field report

Every report on this page started with this scan.

Enter your domain and get the same audit: your AI visibility score, the prompts where competitors get recommended instead of you, and what's causing it.