Building a dental recall system that works

Why recall fails, in one sentence

It is a report, and reports need someone to remember to run them.

That is genuinely the whole diagnosis in most practices. The data exists, the software will produce the list, and on a Tuesday with two emergencies and a staff absence nobody opens it. Nothing about that is a motivation problem. It is a design problem: the work has no state, so it has no owner and no obvious next action.

What a working recall system looks like

Change it from a list into a queue, where every patient carries a state:

• Due — the recall date has arrived, nobody has been contacted yet.
• Contacted — a message went out, on a specific date, through a specific channel.
• Responded — they replied, in either direction.
• Booked — there is an appointment.
• Unreachable — two attempts, no response, moved off the active queue.

The moment recall has states, three things become possible that were not before: you can see what is stuck, a second attempt becomes an obvious action rather than an act of memory, and you can measure the thing rather than argue about it.

The second attempt is where the money is

Almost every practice contacts once. Almost none contact twice.

A single reminder reaches a patient who happens to be looking at their phone in a moment when booking is easy. That is a narrow window, and missing it is not a decision by the patient — it is an accident of timing. A second attempt, a week later, in the same channel, converts a meaningful share of people who fully intended to book and simply did not.

If you change one thing about your recall this month, add the second attempt.

Use the channel the patient answers

Practices tend to standardise on one channel because it is easier to administer. It is also why response rates sit where they do.

Record the channel each patient last responded on, and use that. Some patients reply to a text in ninety seconds and will never answer an unknown number. Some are the reverse. A practice that texts everyone will beat one that calls everyone — and a practice that uses each patient’s own channel beats both, at no extra cost.

Two rules that matter regardless of channel:

• Make booking possible inside the message. A reminder that requires calling back during office hours converts a fraction of one that offers two times and takes a reply.
• Make it recognisable. The practice name in the first four words. An unrecognised number asking someone to book a health appointment reads as a scam, correctly.

What to measure

Not “recall sent”. That is activity, not outcome. Three numbers, monthly:

1. Overdue patients contacted — what share of the due queue actually got a message.
2. Response rate by channel — which tells you whether your channel choice is right.
3. Appointments booked from recall — the only number that pays anyone.

If the first number is below 90%, the problem is the process and no amount of message rewriting will fix it. If the first is high and the third is low, then it is the message.

Where the compliance line sits

Recall is a health communication about an identified patient, which means it carries obligations that a marketing text does not. Keep messages minimal — that a check-up is due, and how to book. Nothing about treatment, findings or history belongs in a text or an email.

This is educational, not legal advice. Your own compliance officer should review the templates, and we build to documented practice rather than telling you the rules are simple.

What we do here

The recall queue is part of the front-desk system rather than a separate product: the states, the timing, the second attempt, the channel memory, and a monthly number for what it booked. What stays with your team is the judgement — which patients need a real call from someone who knows them. The system’s job is to make sure nobody falls off the list, not to replace the person.