A pilot design, with a rehearsed readout on invented cohorts. Not a customer result.
Pilot design
A pilot you can hold us to.
Twelve weeks in one outpatient department. The comparison arms, the population and the measures are agreed before the first message goes out. Reminders alone are the baseline, so the readout has to beat them.
What we compare
Five arms, one department, twelve weeks. Patients are assigned by appointment, and every arm keeps the consent rules on the Trust page.
Baseline
Reminders as they are today
The department's current reminder. This is the baseline everything else has to beat.
Arm 1
Text reminders
The same content by text, on the patient's stated channel.
Arm 2
Guided messages
Confirm, reschedule or ask for help from the message. Travel and parking help the morning of.
Arm 3
Guided messages and wayfinding
Adds the indoor route from the garage to check-in, with location shared only on campus.
Arm 4
Guided messages and follow-up
Adds the scheduling offer when the provider recommends a follow-up.
Who is in it
One outpatient department for the whole pilot. Any of these fit. Radiology is the demo day because its slots are the most expensive to lose.
- Radiology
- Cardiology
- Oncology
- Orthopedics
- Specialty clinic
- Ambulatory surgery intake
What we measure
Seven primary measures, each with a direction and a threshold agreed with the department before launch. We report every one of them, whichever way it moves.
- 1Confirmation rateBetter upThreshold set with you
- 2No-show rateBetter downThreshold set with you
- 3Late arrivalsBetter downThreshold set with you
- 4On-time registrationBetter upThreshold set with you
- 5Reply rateBetter upThreshold set with you
- 6Wayfinding completionBetter upThreshold set with you
- 7Follow-up scheduled within windowBetter upThreshold set with you
Also counted
- Front desk interventions
- Reschedules captured instead of lost
- Arrival predictability by hour
- Patient and staff satisfaction
- Disruption to the department
What we want to learn
- Which patients respond to text and which need a call
- Which moments change behavior and which are noise
- What predicts friction at booking time
- What to scale to a second department
What the desk watches
Nine tiles on the department desk during the pilot. It sits beside the hospital's own analytics and does not replace them.
Department desk · pilot view
At risk of no-show
··Likely late
··On campus, not checked in
··Needing assistance
··Risk by department
Replies by segment
Wayfinding completion
Follow-up completion
Trend by week
The week-six readout, rehearsed
This is the format the department gets at week six and week twelve. The numbers below are invented so the page can show the shape. A real pilot fills them from the record.
Week 6 of 12 · Advanced imaging · rehearsal
- No-show rate
- 4.1%vs 7.9%
- Pilot cohort vs comparison group. Better when lower.
- Late arrivals (over 30 min)
- 9%vs 16%
- Pilot cohort vs comparison group. Better when lower.
- On-time registration
- 81%vs 66%
- Pilot cohort vs comparison group. Better when higher.
- Follow-up scheduled within window
- 58%vs 44%
- Pilot cohort vs comparison group. Better when higher.
Arrival bands
Share of appointments in each band. Colored is the pilot cohort, slate is the comparison group.
Pilot cohort
Comparison group
- Arrived before start: 24% vs 17%
- Checked in on time: 47% vs 40%
- Likely late: 9% vs 14%
- Predicted arrival after start: 6% vs 9%
- On campus, not checked in: 6% vs 7%
- Detected, no kiosk record: 4% vs 5%
- At risk of no-show: 4% vs 8%
We label metrics by what the data proves. Follow-up scheduled is not clinical adherence.
The floor under all of it
- Location only while on campus, and only with opt-in.
- The minimum data to run the arm the patient is in.
- Every send, hold and reply on an auditable log.
- Plain disclosure at booking of what we hold and why.