Every text, judged.
Three jobs your front desk can’t do at scale.
Each one is a judgment call. Each one fails badly when handled by a dumb broadcaster. dentdial runs all three quietly in the background, with a written audit trail you can defend to a CQC inspector or an unhappy patient.
Empty slots, refilled before lunch.
When a patient cancels in Dentally, dentdial pulls your live waitlist within seconds. AI scores every candidate by likelihood-to-accept — looking at preferred clinician, slot type, distance, history of taking short-notice offers — and texts the top match a personal offer. Twenty-minute reply window. No reply, no problem: it rolls down the list automatically. The first YES books straight into Dentally and the chase stops. Average fill time during business hours: under 12 minutes.
Reviews from happy patients only.
After every completed appointment, dentdial decides whether to ask for a Google review — and crucially, when not to. Two-stage safety: a deterministic pre-filter rules out anything that smells like rework, refund, repeat treatment or complaint. Then Claude reads the patient’s last six months and applies a softer sentiment judgment. The result: a 4.9★ profile that actually reflects the practice, not a battlefield of one-star surprises.
Knows a complaint from a thank-you.
Every inbound SMS is classified in seconds. Regex catches the deterministic stuff — STOP, START, YES, NO — and Claude re-classifies the rest into confirm, reschedule, question, complaint or sentiment. Complaints are escalated to a named team member the same day with full context. Ambiguous messages get a drafted reply your staff can send in one tap.
The anatomy of a no.
Most automation tools are proud of what they send. dentdial is proud of what it doesn’t. Here are the signals it watches for — and what each one means in practice.
Red-flag phrases in the notes.
If today’s appointment notes mention any of: rework, redo, remake, refit, recement, fallen out, came out, came loose, failed, fractured, broken, post-op pain, sensitivity since, swelling, not healing, infection, complaint, unhappy, dissatisfied, second opinion, refund, fee waived, FOC, goodwill, ex-gratia — no review request goes out. Full stop. Logged with reason.
A pattern of cancellations.
Three or more cancellations or two or more failed-to-attend appointments in the last six months. A patient who keeps cancelling is rarely the patient you want endorsing you publicly. Skipped.
Same treatment, ninety days.
If the same treatment was performed in the last 90 days, that’s a rework cycle signal. Even if no one wrote it down. Skipped.
Prior complaint history.
If this patient’s number has ever sent an inbound message classified as a complaint in the last six months, no review ask. Ever.
AI sentiment check.
If the deterministic filter passes, Claude reads the full appointment notes and patient history. It applies softer judgment — was the clinician’s language about this patient warm or guarded? Were there hints of frustration? Did the patient mention something that didn’t go to plan? If anything is off, the AI returns should_send: false with a written reason logged to the audit trail.
A cancellation, refilled in real time.
This is what happens between a patient hanging up at 09:03 on a Tuesday and someone else sitting in that chair at 11:00 the same day.
What it’s worth, in plain numbers.
Two metrics that move on the very first week — and two that move slowly but matter more.
£3k–£18k / month recovered.
A typical two-clinician practice runs 15–30 cancellations a month. At an average slot value of £180 across routine and complex, that’s £2,700–£5,400 of pure recovery on routine alone. For a busy implant or cosmetic list, the upper bound stretches well into five figures.
2–4 hours / day saved.
Hours your reception team spent chasing the waitlist, drafting review-ask texts and triaging inbound replies — handed back. Used by most practices to focus on the patients in front of them.
+0.3 stars on Google.
The compounding effect of asking only happy patients for reviews. Rough benchmark from our own data: practices move from 4.5–4.6★ to 4.8–4.9★ within six months, even with no other change. Reviews drive new-patient enquiries; new-patient enquiries drive everything else.
Complaints handled in hours, not weeks.
Patient frustration caught at the first SMS reply is patient frustration that doesn’t become a one-star review, a GDC referral or a refund. dentdial routes it to the right human before it grows.
A pure Twilio reseller can’t differentiate. A system that reads complaints in real time, drafts the reply, and knows when not to ask for a review — can.
— Internal product note · April 2026
SMS automation is a commodity. Judgment isn’t. Every message dentdial sends carries a written record of why it was sent — and every message it doesn’t send carries a written record of why it didn’t. That’s the audit trail that turns automation from a liability into a defensible asset.