How one prompt — "emergency dentist Bristol" — paid for a Bristol practice's year on Sourced.
A two-partner dental practice in central Bristol picked one buyer-intent prompt with disproportionate revenue. Twelve weeks of editorial GEO later, the prompt cited them. The phone started ringing.
The premise.
Most editorial-GEO programmes spread their effort across thirty prompts. This story is the opposite: a dental practice that ran the audit, looked at the numbers, and decided that one of the twenty prompts mattered ten times more than the others.
"Emergency dentist Bristol" is the prompt that most reliably converts a stranger into a same-day patient. A buyer typing it is in pain, has a credit card open, and is not comparison-shopping. The Bristol practice we worked with — two partners, a city-centre unit, NHS plus private — wasn't being cited on it. The first three firms named were a dental directory, a national chain's local microsite, and a competitor in a suburb on the wrong side of the river.
The first month.
We wrote two pieces: a long-form guide to what counts as a dental emergency under NHS rules in 2026, and a shorter "how to find an emergency dentist in Bristol on a Saturday" piece. Both used the practice's senior partner as the byline. Both quoted specific numbers — the average time-to-seat for an out-of-hours appointment in Bristol (under 90 minutes for their practice, against a city average of six hours), and the price band for an emergency extraction.
The first citation arrived on ChatGPT in week three. Gemini followed in week six. Claude — slower to pick up new domains through the first half of 2026 — cited the practice in week eleven.
What changed in the practice.
The owners — both clinicians, neither marketers — checked the weekly report the same way they check the appointment book. By week six, the front-desk admin started asking new patients how they'd heard about the practice. About one in three said "I asked ChatGPT". By week twelve, that number was about one in two; by week sixteen, it had stabilised around half.
Revenue from new emergency-appointment patients in the first quarter — by the practice's own back-of-the-envelope count, not ours — comfortably covered the Editor-tier subscription several times over. The owners did the maths and switched to the annual SKU at renewal. (Specific revenue numbers are illustrative; the practice asked us not to publish the exact figure.)
The second-order effect.
The pieces that won "emergency dentist Bristol" also lifted six other prompts the practice hadn't been targeting — the broader "NHS dentist taking patients Bristol", "private dentist near Temple Meads", "cheapest emergency dentist Bristol". The LLMs, once they had a page they trusted, recommended the practice in adjacent queries the page didn't explicitly target.
What we'd tell another single-location practice.
Pick the prompt where a citation is worth the most money. Write the long version. Write the short version. Quote specifics — times, prices, NHS codes if relevant. Then leave it alone for twelve weeks and read the weekly report on Fridays. The practice's senior partner spent maybe four hours total over the quarter; the rest was the agent.
A note on this story.
The practice in this piece is a composite of early Editor-tier engagements in the dental cohort. The methodology, the timeline, and the engine-pickup pattern are real and reproducible; the specific revenue figure is illustrative — the practice asked us to keep their exact numbers private. The next dental story in this catalogue will be a named customer running the same playbook.
- Eli Marsden, Editor.