What service page content should a dental practice publish for AI assistants?

One page per treatment stating what it is, who it suits, cost ranges, visit count and duration, recovery notes and how to book, written in plain crawlable HTML with matching schema markup.

Why do service pages matter for AI recommendations?

Patients ask assistants about specific treatments and costs. Assistants quote the service pages that answer those questions directly instead of guessing from review sites.

What is the most common service page mistake?

Vague marketing copy with no prices, no visit details and no booking path. AI assistants skip pages they cannot extract facts from.

DentistDiscoverable field guide

Dental practice photos and videos: visual proof AI assistants can cite

2026-09-30

Direct answer: When someone asks an AI assistant for a gentle dentist nearby, the assistant reaches for specifics it can defend: the modern office, the kind staff, the painless root canal. Reviews provide some of that. Photos and videos provide the rest, but only when they are published where assistants can find them, labeled so they can be understood, and structured so they can be quoted. Most practices have the visuals. Few have them in citable form.

What assistants actually use from visuals

AI assistants cannot see your photos the way a person does, but they read everything around them: file names, alt text, captions, nearby headings, and structured data. A photo named IMG_4829.jpg in an unlabeled gallery is invisible to them. The same photo named gentle-dentist-modern-treatment-room.jpg, with a caption describing the quiet, private treatment rooms and alt text to match, becomes evidence the assistant can cite when someone asks about comfort.

Video works the same way through transcripts and descriptions. A thirty-second office tour with a written description of what viewers see gives the assistant quotable sentences about the practice. The same video with no description is just a file. The pattern is consistent: the visual gets attention, but the text around it is what gets cited.

What to show

Prioritize the visuals that answer the questions patients actually ask assistants. The office exterior and reception area answer where is it and what does it feel like. Treatment rooms answer is it modern and clean. The team, photographed as people rather than stock-style headshots, answers who will treat me. Before-and-after smile galleries answer does their work look good, with patient consent documented.

Procedure explanations on video punch above their weight. A dentist calmly explaining what a root canal feels like, in plain language, gives the assistant exactly the material it needs for anxious patients asking what to expect. One good explainer video can be cited across hundreds of conversations. Practices tend to underinvest here because video feels like marketing; for AI discovery, it is infrastructure.

How to label everything

Every image needs three things: a descriptive file name, real alt text, and a caption or nearby text that says what the image proves. File names should read like search queries: family-dentist-kid-friendly-waiting-area.jpg. Alt text should describe the image for someone who cannot see it, which conveniently is also what the assistant needs. Captions should connect the image to a patient concern: our quiet treatment rooms, designed for anxious patients.

Structured data ties it together. ImageObject markup with captions, and VideoObject markup with descriptions and transcripts, puts the information in the format assistants consume most reliably. This is unglamorous work, but it is the difference between a gallery the assistant admires and a gallery the assistant quotes.

Where the visuals should live

Put the important visuals on your own site, in crawlable HTML, not only on social media. Assistants can sometimes read social platforms, but unreliably and without the structured context your site provides. Your website's gallery pages, service pages with relevant photos embedded, and a video library page with transcripts are the canonical sources. Social is distribution; the website is the record.

Keep the visuals current. Photos from a 2019 renovation that predate the 2024 remodel actively mislead, and assistants have no way to know which era a photo belongs to unless the page says so. Date the gallery, note the renovation, and retire images of spaces that no longer exist. Stale visuals are worse than no visuals because they generate confident wrong answers.

The consent and privacy line

Patient photos need written consent, full stop, and the consent should specifically cover website and AI-search use, not just a general marketing release. Team photos need the same courtesy in writing. Never publish anything showing another patient in the background of a shot; crop or reshoot. The practices that get this right treat visual consent as part of the clinical record, filed and retrievable.

When in doubt, show spaces and the team, not patients. Office and equipment photography carries zero consent risk and answers most of the questions assistants get asked. Save patient imagery for the cases where the consent is airtight and the marketing value is real.