Anonymised. Drawn from two regional allied-health programmes.
Background
A physiotherapy practice in Ballarat (three practitioners, established 2016, bulk-billing and private-fee services) was invisible in AI-generated answers for “physiotherapist in Ballarat” and its variants across all four major AI systems. The practice had a functional website, 140+ Google reviews averaging 4.8 stars, and a reasonable local SEO ranking. Visibility Index at audit: 16 out of 100.
What the audit revealed
Three structural problems, each fixable. No schema on the website: the WordPress site had no MedicalBusiness or Physician schema; the models had no structured signal confirming the business type, the practitioners credentials, or the conditions treated. One model was categorising the practice as a “wellness centre” rather than a physiotherapy clinic, which affected which queries it appeared in.
Practitioner profiles were non-existent. In allied health, the practitioners are the entity as much as the practice. A physiotherapy clinic with no structured profiles for its physios (no AHPRA registration numbers in schema, no authored content, no Person schema) is invisible in the “best physio for condition” queries that account for roughly 60% of the category high-intent search volume.
Third-party mentions were thin and generic: two health directories, both with generic entries that mentioned “physiotherapy” but no specific conditions, techniques, or service differentiators.
The programme; 11 weeks
Weeks 1 to 2: Schema foundation. Added MedicalBusiness schema with medicalSpecialty, availableService, and priceRange. Added Person schema for all three practitioners with AHPRA registration numbers, credential lists, and sameAs links to their AHPRA public register entries. This alone moved the index from 16 to 34.
Weeks 2 to 5: Condition-specific content. Published six condition pages (lower back pain, sports injuries, post-surgical rehab, neck pain, pelvic floor, dry needling) each written with a specific structure: condition definition, how physiotherapy addresses it, what to expect in sessions, approximate duration of treatment, and explicit FAQ schema addressing the questions buyers ask AI assistants. Average length: 1,600 words. Each page authored and schema-attributed to a specific practitioner.
Weeks 4 to 8: Citation engineering. Placed practitioner profiles in the Australian Physiotherapy Association member directory (previously unlisted). Secured two editorial placements in the Ballarat Courier health section. Earned a mention in a local community Facebook group thread about post-natal physio. Added to the Better Health Channel regional provider search.
Weeks 6 to 11: Review structure. Added AggregateRating schema to the homepage pulling from Google review data. Launched a structured post-appointment review request SMS that directed patients to leave condition-specific reviews mentioning the treatment type; not just stars, but context. “Great help with my back pain after the marathon” is a more useful retrieval signal than “5 stars, highly recommend”.
Results at week 11
- Visibility Index: 16 to 71
- ChatGPT: named in 9 of 12 priority queries (from 0)
- Perplexity: named in 11 of 12 priority queries (from 2)
- New patient enquiries via website: up 41% week-on-week in weeks 9 to 11 versus prior period
- Specific condition queries (“physio for lower back pain Ballarat”): top recommendation in all four models
The learning
Allied health is a category where practitioner-level entity work unlocks category-level visibility. Adding AHPRA numbers to Person schema sounds like a minor detail; it produced an 18-point index jump in two weeks because it is a verifiable, authoritative credential signal that the models treat with high trust. In any licensed profession, the equivalent move (licence number plus credential body sameAs) is the highest-ROI single action available.