Course completion is the economic unit
A physiotherapy or allied health business sells courses of treatment rather than single appointments. A patient who stops after two of six sessions represents lost revenue and, more importantly, an incomplete outcome.
Predicting who is at risk of dropping out, early enough to do something, is the most valuable operational application - and the intervention is usually a conversation rather than anything technical.
What predicts drop-out
- Gap between booking and first appointment
- Whether the second appointment was booked at the first
- Distance travelled, and appointment time relative to working hours
- Whether the patient is self-funding or covered by insurance or an employer
- Early cancellation or rescheduling behaviour
- Referral source, which correlates with commitment more than expected
The second point is usually the strongest single signal. A patient who leaves without a next appointment booked is substantially less likely to return, which makes booking at the point of care a concrete, cheap intervention.
Filling cancelled slots
Short-notice cancellations are a direct loss in a business where capacity is clinician hours. Predicting which appointments are likely to cancel allows a standby list to be prepared rather than assembled reactively.
Predicting which patients would accept a short-notice offer is equally useful, and less commonly done. Contacting the twenty most likely to accept beats contacting everyone and irritating the rest.
Referral source analysis
| Metric by source | What it reveals |
|---|---|
| Referrals received | Volume - usually all that is tracked |
| Conversion to first appointment | Whether referrals are appropriate |
| Course completion rate | Quality of the referral match |
| Revenue per referral | The number that actually matters |
| Repeat or onward referral | Long-term relationship value |
Most clinics track only the first row. Analysing the rest frequently changes where relationship effort is directed, and it is straightforward from data already in the practice system.
Keep clinical decisions clinical
Predicting treatment outcomes, recommending protocols or triaging clinical urgency are regulated activities with a much higher evidence bar and real patient safety implications.
An operational analytics project should not drift into them. If clinical decision support is genuinely the goal, it needs clinical governance, appropriate validation and regulatory advice from the start - a different project with a different shape.
The patient who leaves without booking the next session is the one to call tomorrow.