How many recall touches before we give up on an overdue patient?

Most practices' honest answer is "one, maybe" — a due-date text from the software, or a postcard, and then the patient either responds or joins the quiet file. Framed as a system design question, nobody would design it that way. It exists because each extra touch feels like pestering. The measured data says otherwise.

The coin-flip problem

In a live practice we measure, single attempts — one callback, one follow-up — behaved like coin flips against busy lives: the message lands during the school run, gets mentally filed, evaporates. Meanwhile the same measured data showed the surprising ceiling: leads everyone had written off closed at 67% once someone actually engaged them. The overdue patient who ignored March's postcard isn't declining care. They missed a postcard. Those are different facts with different price tags.

Three touches, spaced, in different channels

The working shape: touch one on the due date (text, one-tap rebook). Touch two ten days later, different angle ("mornings or afternoons work better?") — a question outperforms a reminder, because questions get answers. Touch three a month out, the honest last word: "We'll stop nudging after this one — whenever you're ready, this link always works." Then the sequence ENDS, on schedule, as promised. Three touches over six weeks is persistence. Twelve touches forever is spam. The line is a published stopping point.

The opt-out is load-bearing

Every touch carries a real way out, and the practice honors it instantly. This isn't compliance theater — it's what makes the whole system sustainable: the list stays clean, replies stay warm, and the patients still on it are there by consent. A recall list people trust is a compounding asset; one they resent is a liability with postage.

Score your persistence

The free score measures all four recall axes — speed, persistence, program, channels — in three minutes, scored 0–100 with your weakest named first and the quiet file's monthly value estimated from your own numbers, assumptions labeled.