AI patient recall for opticians, and how the automation actually runs

Why recall is where the money sits

A patient who is overdue for a test generates nothing until they come back through the door, and every month they stay overdue is a month a competitor might see them instead. Recall automation exists to stop that drift being invisible: instead of a paper list nobody owns, an overdue patient enters a queue automatically and stays in it, worked, until a person or a clear rule takes them out. [FILL: Winstanley & Son's own recovered-recall numbers, once a full campaign cycle has run and been measured, for use as the case-study figures]

How the power dial actually works

Patients enter a "Power Dial" calling queue automatically thirty days before their recall falls due. Once in, a patient is re-queued for another attempt every three days, on a loop, forever. There are only two ways out: a member of staff moves that patient's card, for example to "will call in", "don't want to be tested" or "gone elsewhere", or a safety condition applies, such as the patient having already booked, or being marked do-not-contact. Nobody has to remember to keep dialling; the list simply never runs dry until a person or a rule closes it.

The same logic covers households that share one phone number: rather than dialling the same number five times for five family members, one record carries the whole household's recall dates and reference numbers, so a call still reaches everyone who is due without repeating the household.

The SMS reactivation drip, for older lapses

For patients who lapsed three years or more ago, a phone call is often the wrong first touch. Winstanley & Son built a text-message campaign instead, deliberately slow: batches of 30 messages a day, in working hours, so the practice's messaging number is never at risk of being flagged for sending too much too fast. Each message names the practice, uses the patient's first name, and always carries an opt-out. A message alternates between a health-led reason to come back and an eyewear offer, and anyone who doesn't respond to the first gets the other angle at the second touch, never the same message twice. If the rate of opt-outs on any single day passes a set threshold, the drip pauses on its own and the whole thing is reviewed before it resumes.

A reply is picked up by the same patient-texting agent that handles day-to-day bookings, so a patient who texts back "mornings" is offered a real appointment slot there and then, including out of hours, rather than waiting for the practice to reopen.

What a practice needs in place before recall automation works

None of this runs on hope. It depends on a small number of things being true in your data before the first patient is ever contacted: a reliable recall-due date on every patient record, a working mobile or landline number, and an accurate do-not-contact and DND list carried over from however the practice currently manages consent. Shared household numbers need handling deliberately, or the same family gets contacted five times over. And because this reaches real patients, every list is checked twice, once when it's built and again immediately before anything sends, against the practice's existing stop and exclusion records.

Questions practice owners ask