Skip to main content
CareConcierge

Essay · Commercial performance

AI May Recommend Your Practice. Then Your Practice Has to Answer.

Principals are being sold AEO and GEO visibility, and visibility matters. But the first useful response shapes whether an introduction becomes a genuine conversation. Here is a practical way for an owner to trace that journey from recommendation to response.

A close-up of an eye seen through an optician’s letter chart.
In this essay
  1. The proposal on your desk
  2. The acronyms on the table
  3. A recommendation is borrowed authority
  4. The Recommendation-to-Response Audit
  5. Three jobs, one owner
  6. Where CareConcierge fits
  7. Back to Thursday

The proposal on your desk

Imagine it is Thursday afternoon and the proposal is on your desk. An agency has been running test questions through AI-assisted search tools, and the screenshots are persuasive. Ask about clear aligners for adults in your suburb, or recovery after a breast reduction in your city, and your practice appears in the answer, cited, linked and described in terms you would happily have written yourself. The proposal recommends a monthly retainer to keep it that way.

Before you sign, you do something no visibility report measures. You open your own website on your phone and send the practice a simple question through its enquiry form, the way a stranger would. Is the first appointment an assessment or a sales conversation? Is there a Saturday clinic? You add a mobile number, note that you would rather be texted than called, and press send at ten to six, as the front desk is closing.

The automatic receipt arrives in seconds. Thank you for your enquiry; a member of our team will be in touch.

The useful answer does not arrive that evening. It arrives mid-morning the next day, by email, asking you to call reception. Neither question is answered. The note about texting has vanished somewhere between the form and the inbox.

The search introduced your practice in seconds. Your practice took most of a day to reply and told the enquirer less than the search had. Nobody in the building did anything wrong. That is the problem.

The acronyms on the table

Answer engine optimisation and generative engine optimisation, AEO and GEO, are the names the marketing industry now gives to the work of appearing inside AI-generated answers rather than only in a list of links. The instinct behind them is sound. If more patients begin their research by asking a machine a question, a practice should care whether its expertise informs the answer.

For its own search features, Google has been unusually plain. Its guidance on AI Overviews and AI Mode treats AEO and GEO as labels you may encounter, and says that from Google Search’s perspective, optimising for generative AI search is still SEO. It tells site owners they can ignore much of the folklore, including llms.txt files, chopping pages into chunks and special schema, and that unique, expert, people-first content matters most over time. It also reminds owners that meeting every requirement does not guarantee a page will be shown. That advice governs Google’s features, not every AI tool a patient might use, but its direction is sensible.

A word on language. These systems rarely say “I recommend this practice”. They cite it, surface it and summarise it. To a patient, though, prominence inside a confident answer can feel like an introduction that carries authority, and that feeling is what the practice inherits the moment she makes contact.

So here is the challenge. Before you spend heavily to win the recommendation, prove that your practice can receive it. Every acronym in that proposal describes the moments before someone makes contact. Not one describes what happens once they do. That part belongs to you.

A recommendation is borrowed authority

When a search experience surfaces your practice, it lends you credibility you have not yet earned with that person, borrowed from her trust in the tool and in the sources it cites. The first conversation is where the practice proves it deserves that authority. If the reply is slow, generic or forgetful, the loan is called in quietly, and nobody sends the practice a statement.

It helps to see the journey as a chain rather than a funnel, because a chain shows exactly where it breaks.

  1. Discovery. She finds you, by search, referral, social media or a friend.
  2. A credible answer. Your content addresses her question honestly, in your own expert voice.
  3. Enquiry. She chooses a channel and tells you what she wants to know.
  4. A useful, bounded response. Her actual question is answered from information the practice has approved, and anything clinical goes to a clinician.
  5. A warm human handoff. A team member receives her context, not just her phone number.
  6. An appropriate next step. A consultation, a callback at the time she chose, or a considered “not yet”.
  7. A measured outcome. Someone records what happened, so the owner can see it.

Many practices invest heavily in the first two links and count the third. The middle of the chain, where expressed intent is either received or mislaid, tends to go unwatched. A form receipt, a missed-call alert and an understood conversation all register as “an enquiry came in”. Only the last tells your team what the person wanted, what she has already been told and how she would like to be contacted.

The Recommendation-to-Response Audit

You do not need a consultant to find where your own chain breaks. You need a fortnight of records, an honest notebook and an hour of your practice manager’s time. I call it the Recommendation-to-Response Audit, and it applies the Reaching and Converting disciplines of The Exit Discipline to a single fortnight. Run it before you sign anything.

Choose the sample. Take the enquiries the practice actually received over a recent fortnight, across every channel you use: phone, web forms, email, WhatsApp, web chat and Instagram. Include those that arrived after hours and at weekends. Work from your own records under your existing privacy obligations; this is an internal review, not material for a case study. Then add a few test enquiries of your own, non-clinical and sent at different hours, as you did on Thursday.

Record five measures for each enquiry.

  1. Entry. Where and when did the enquiry arrive?
  2. Useful response. How long until the person received an answer that did more than acknowledge receipt?
  3. Question answered. Did the response address what was actually asked, within approved boundaries?
  4. Context preserved. Did the next human receive the person’s question, preferences and prior exchange?
  5. Next step owned. Was an appropriate next action documented, with a responsible person or pathway?

Count honestly. Your denominator is the enquiries you received. It is not the people who saw your page and never wrote, and it is not a market estimate. If you cannot count received enquiries reliably across every channel, that is your first finding, and possibly the most important one.

Treat speed as necessary, not sufficient. An automated receipt is not a response. A fast wrong answer is not service. Speed is one column in the notebook, not the verdict.

Check the clinical boundary. Where an enquiry contained a clinical, urgent or ambiguous question, note whether it reached an appropriate clinician through a pathway the practice designed.

Regulatory detail differs between professions and jurisdictions. The operating discipline does not: the practice approves what may be said, patient-specific clinical questions follow the appropriate human pathway, and speed never outruns accuracy.

When the notebook is full, look for the link that breaks most often. In my experience it is rarely the clinician and rarely the coordinator’s goodwill. It is usually the handoff, where context is lost because the system was built to collect contact details rather than to understand a person.

Three jobs, one owner

The audit usually reveals that one asset has been asked to do three different jobs.

Your website and its expert-led content exist to make the practice discoverable and credible. That is where search work, including whatever the AEO conversation becomes, properly lives.

Your communication system exists to protect the moment of expressed intent, when a person stops browsing and starts asking. Its job is to receive, to answer within approved bounds, to keep context and to route the conversation to the right human.

Your team does the human work no system should pretend to do: judgement, explanation, an informed-consent process led by appropriately qualified practitioners, and care.

The owner governs all three. A principal who sees the chain whole can design the journey deliberately, instead of buying another fashionable acronym and hoping a result arrives on its own.

Where CareConcierge fits

The Exit Discipline treats this as the connection between two of its five levers, Reaching and Converting. Reach that cannot be received is incomplete; conversion that cannot be measured is easily romanticised.

CareConcierge works at that join, and only there. It does not influence search visibility or AI recommendations. Its work begins when someone writes.

Across the channels a practice configures, it provides AI-assisted responses drawn from practice-approved content. It gathers non-clinical intent and preferences, follows escalation rules the practice defines for anything clinical, urgent or out of scope, and prepares a warm handoff so a team member continues with context.

For the owner, the monthly Conversion Intelligence Report is designed to show where enquiries progressed or stalled, which questions recur, and where booking evidence is missing or not yet validated by the practice. What any practice sees depends on its configuration, its team’s workflow and the enquiries that actually arrive. The previous essay in this series set out why that bounded role is the right one.

Back to Thursday

Run Thursday again, in a practice that has done the audit.

At ten to six you send the same two questions. Within minutes a reply arrives on the channel you used. Drawing on the practice’s approved information, it explains that the first appointment is an assessment with the clinician, what that involves and how fees work, and it lists the published Saturday sessions. It notes that questions about your own situation are for the clinician at that assessment. Then it asks when you would like the team to be in touch, and you say a text in the morning.

At nine the next morning the coordinator opens a short briefing: what was asked, what was answered, that you prefer text and that Saturdays matter. The message that follows reads like the second conversation, not the first.

That is the standard. Not a promise that every enquirer books, because many will rightly take their time, but proof that the practice can receive what the search sends it.

So run the test today. Pick up your phone, choose one channel, ask one plain question and time what comes back. Then decide what to spend.

The future of search may decide who is introduced. What becomes of the introduction is still yours to decide.

Gregory William Gray has spent more than two decades advising owner-led dental, orthodontic and plastic surgery practices in Australia, the United Kingdom and the United States. He is the author of The Exit Discipline and co-founder of CareConcierge Health. More essays and videos are in Inside Private Practice.

Source notes

All sources accessed 27 October 2026. Google’s guidance applies to Google Search’s own AI features; regulatory references support the operating principle in the audit and do not imply that every one-to-one enquiry response is advertising under every regime.

  1. Google Search Central, Optimizing your website for generative AI features on Google Search (last updated 10 July 2026).
  2. Google Search Central, AI features and your website (last updated 10 December 2025).
  3. Ahpra and National Boards, Guidelines for advertising a regulated health service (effective 14 December 2020).
  4. Ahpra, Meeting your professional obligations when using Artificial Intelligence in healthcare (reviewed 22 August 2024).
  5. General Dental Council, Guidance on advertising (effective 30 September 2013).
  6. General Medical Council, Cosmetic interventions: key points.

A 30-minute video conversation with a founder about how enquiries reach your practice, where conversations stall and what your team needs next. We will show how CareConcierge could fit your workflow. If it is not a fit, we will say so.

Contact CareConcierge

How would you like to continue?

Choose whichever suits you. A small question is welcome; there is no need to book anything first.

Ready for a proper conversation?

A 30-minute video conversation with a founder about how enquiries reach your practice, where conversations stall and what your team needs next. We will show how CareConcierge could fit your workflow. If it is not a fit, we will say so.

Discuss your practice

Please do not include patient-identifiable or clinical information.

Ask a question Discuss your practice