Dental Recall Automation UK: Where It Leaks, and Why

Dental recall automation in the UK works when it automates the chase and leaves the interval alone. The reminder, the escalation and the record of both are mechanical, but the recall interval itself is a clinical judgement that NICE guideline CG19 says must be set for the individual patient — 3 to 24 months for adults, 3 to 12 months for under-18s — so a system that applies one default period to every patient has automated the wrong half of the process.

Recall is where the revenue sits in a dental or optical practice, and it is almost absent from the automation category. Search for recall software and you get platforms describing their own reminder modules; the two questions that actually decide whether recall works — what interval, and what you are allowed to send — go unanswered. This piece is the diagnostic, not a product comparison.

What is dental recall automation?

Dental recall automation is software that tracks when each patient is next due, contacts them without anyone typing the message, escalates when there is no reply, and records what was sent. It is distinct from appointment reminders, which is where the confusion starts. A reminder chases an appointment that already exists in the diary, days out. A recall chases an appointment that does not exist yet, months out. They share a channel and nothing else.

The same distinction holds in optical practice, where recall drives sight tests and the dispensing that follows. Both verticals have well-developed reminder tooling and neglected recall tooling, because a missed reminder produces a visible empty slot today, while a failed recall produces nothing at all — a patient who never comes back, and never appears in a report.

Why is the recall interval the one thing you must not hardcode?

Because in dentistry it is a clinical decision with published guidance behind it. NICE CG19 sets recall intervals of 3, 6, 9 or 12 months for patients under 18, and 3, 6, 9, 12, 15, 18, 21 or 24 months for patients aged 18 and over, with the shortest interval for any patient being 3 months. Intervals may be extended over time, up to 24 months, for patients who have repeatedly shown they can maintain oral health and are not considered at risk (NICE CG19 recommendations, checked 18 September 2026).

Read that as an operations problem rather than a clinical one and the implication is blunt: the interval is a per-patient field that a clinician sets, and automation's job is to act on whatever is in that field. A recall workflow that says "six months after the last visit, send the text" is not automating admin. It is overwriting a clinical judgement with a default, and it does it silently, to the whole list at once.

The optical equivalent is subtler and catches more practices. The commonly quoted two-year figure is a funding interval, not a clinical recall interval: NHS guidance is that an eligible patient can have a free NHS sight test every two years, and more often where the optometrist recommends it (NHS, How often can I have a free NHS sight test?, checked 18 September 2026). FODO makes the same point from the profession's side — the frequencies in the Memorandum of Understanding are periods that govern how a General Ophthalmic Services form is coded, not recommended sight-testing intervals (FODO, sight test/eye examination intervals, checked 18 September 2026). The College of Optometrists frames frequency as depending on clinical need, noting for example that annual examination may be appropriate for myopic children or those at risk of myopia until around 12 to 13 years old, then every two years (College of Optometrists, frequency of eye examinations, checked 18 September 2026).

So a practice that set its recall default to 24 months because "the NHS says two years" has encoded a reimbursement rule as a clinical one. Contact lens wearers, diabetic patients, glaucoma suspects and children are the population that default silently under-serves.

None of this is clinical or legal advice: the interval is your clinician's call, and the guidance above is the source to read rather than a substitute for your own adviser.

Where does recall actually leak? A diagnostic sequence

Run this in order over your last twelve months. Stop at the first step that fails, because fixing a later step while an earlier one is broken changes nothing.

1. Does every active patient have a recall date at all? Pull the count of patients with no next-due date recorded. This is the leak nobody expects and it is usually the largest one. A patient with no recall date is not overdue in any report — they are invisible.
2. Is the date a clinician set, or a default the system applied? Sample fifty records and check whether the interval varies. If nearly every adult sits at six months, you have a default, not a judgement. That is the finding this whole post exists for.
3. Is there a contact route on file that works? Count records missing a mobile number, missing an email, or carrying a marketing preference that blocks the channel you plan to use. A recall that cannot be delivered is not a recall failure, it is a data failure.
4. Does the system stop chasing when the patient books? Test it. Book a test patient mid-sequence and see whether the remaining messages still go out. A sequence that keeps chasing someone who has already booked costs you more credibility than the appointment was worth.
5. What happens on no reply? Write down the actual escalation: second message, channel change, a call from a human, or nothing. "Nothing" is the honest answer in most practices, and it is where the recoverable revenue is.
6. When does a patient stop being overdue and become lapsed? If there is no defined threshold, there is no reactivation process either — just a list that grows.
7. Can you show what was sent, to whom, when, and on what basis? If the answer is a phone screen rather than a log, you do not have a recall system. You have someone remembering.

Steps 1 to 3 are data work and cost nothing but attention. Steps 4 to 7 are where automation earns its keep.

When does a recall reminder become marketing?

This is the line that decides what you may send, and it falls inside the recall sequence rather than around it. The ICO treats a factual service message — including a message that reminds someone about an appointment — as outside the direct marketing rules, but it is equally clear that a message carrying promotional elements counts as direct marketing even when that is not its main purpose (ICO, Identify direct marketing, checked 18 September 2026).

Where a message is direct marketing sent by electronic mail, regulation 22 of PECR requires consent, or the existing-customer soft opt-in: contact details obtained directly from that person, marketing limited to similar goods or services, and a clear opt-out offered both when the details were collected and in every message since (ICO, Electronic mail marketing, checked 18 September 2026). We set this out at length for intake and rebooking in therapy practice automation; recall is the same rule at a different point in the patient lifecycle.

In practice that means three message types, and most practices run them as one. "You are due for a check-up, call us on 01234 567890 to book" is a service message. "It has been a while — book this month and save £20 on a hygiene visit" is direct marketing, whatever heading it sits under in your software. And a win-back to someone who last attended three years ago is direct marketing to a person whose soft opt-in has probably gone stale, on the basis that recency is part of what makes the exemption defensible.

The awkward corollary: the further overdue a patient is, the more tempting promotional wording becomes and the weaker your footing gets. Keep the sequence factual as long as you can, and treat reactivation as a separate campaign with its own consent check.

Should a recall message name the treatment?

Default to no. The moment a message links an identifiable person to a health service, you are processing data concerning health, which Article 9 of the UK GDPR treats as special category data and holds to a higher standard (ICO, Special category data, checked 18 September 2026). "Your hygiene appointment is due" is mild; naming a procedure, a condition or a device on a message that may be read on a lock screen by someone other than the patient is a disclosure risk you get nothing for. Keep clinical detail out of the channel and inside the record.

If you are introducing AI into the sequence rather than plain templated messaging, assess whether you need a data protection impact assessment before you switch it on, not after.

What does your practice software already do?

More than most practices use. Recall handling is long-established in the UK dental systems — Software of Excellence EXACT, Carestream R4 and Dentally all ship it, and Dentally exposes a documented API for integration. UK optical systems such as Raven Vision and Perception include recall as a core module. Check four things before commissioning anything:

• Whether the recall interval is a per-patient field your clinicians can actually set at chairside, and whether they do.
• Whether the built-in sequence supports escalation and channel switching, or only a single send.
• Whether it stops on booking, and whether it logs every send against the patient record.
• Whether marketing preference is stored separately from contact details, so a reactivation campaign can respect it without a manual filter.

If all four are already there, the right answer is configuration and a data clean-up, not a new system. The case for a custom build starts where your software stops: joining recall data to something it does not talk to, or an escalation path that needs judgement applied to a list. That, and not the reminder itself, is the work worth paying for — the same conclusion we reached about reducing no-shows at a UK clinic, where the cheapest fixes were the ones already sitting inside existing tools.

Frequently asked questions

What is dental recall automation? Software that tracks when each patient is next due for an oral health review, contacts them without anyone typing the message, escalates when there is no reply, and logs what was sent. It differs from appointment reminders, which chase an appointment already in the diary. Recall chases an appointment that does not exist yet, months out.

Can you automate the recall interval itself? No, and this is the central point. NICE guideline CG19 sets the interval as a per-patient clinical judgement — 3 to 12 months under 18, 3 to 24 months for adults. Automation should read whatever interval the clinician recorded and act on it. A workflow that applies one default period to the whole list has replaced a clinical decision with a setting.

Is a dental recall reminder direct marketing under UK rules? A factual reminder that a check-up is due is a service message and sits outside the direct marketing rules. The ICO is clear that a message carrying promotional elements counts as direct marketing even when that is not its main purpose, so adding a discount or an offer to the same text moves it across the line and brings PECR regulation 22 into play.

Can I text a patient who lapsed three years ago? Carefully, and not with an offer by default. A promotional win-back is direct marketing, so it needs consent or the soft opt-in, which requires details you collected directly, marketing about similar services, and an opt-out in every message. Recency is part of what makes that exemption defensible, so a three-year gap weakens it.

Should a recall text name the treatment? Usually not. Linking an identifiable patient to a named procedure or condition is processing data concerning health, which Article 9 of the UK GDPR treats as special category data. A message may be read on a lock screen by someone other than the patient, so keep clinical detail in the record and out of the channel.

Does an optician's two-year recall come from the NHS rules? Not as a clinical interval. NHS guidance is that an eligible patient can have a free sight test every two years, and more often where the optometrist recommends it, and FODO notes that the Memorandum of Understanding frequencies govern how a General Ophthalmic Services form is coded rather than setting recommended testing intervals.

Where to start

Run the seven-step diagnostic before you look at software. If steps 1 to 3 fail you have a data problem, and no platform fixes a patient list with no recall dates on it. If steps 4 to 7 fail, the gap is workflow, and that is where a build pays for itself.

Ihsan Ops is a UK AI automation agency in Bedford that builds this kind of workflow for appointment-based practices. Every engagement starts with an AI Opportunity Audit from £1,000, which is how scope and price get established. See what that covers under process automation, or book a 30-minute discovery call and we will tell you honestly whether your existing system already does the job.