Returned on low-attendance warning
Caught before the drop, 265 of 295
Of 7-day silent quitters returned
Within two weeks, 88 of 135 caught
AI drafts sent without an edit
Over 6,800 drafts, sent word-for-word
Coach time back per week
Context-gathering and first draft done for him
The problem
When success runs through one phone, the quiet members drift.
Coach Mike runs a 222-member personal-training studio in the UK. Before us, every check-in, every no-show follow-up, every onboarding sequence ran through his phone.
Which meant the work always got done for the loudest members. The quiet ones drifted. One missed week, then two, then a polite "I'm going to pause" message six weeks later.
"Every Sunday I'd sit down and WhatsApp every member a check-in. By the time I'd finished, half my day was gone. And that was just the check-ins."
Who has gone quiet
Where every active member sits right now, by days since last session. Without the system, every name in the bottom half is one Coach Mike would have to remember on his own.
| Days since last session | Members | What it means |
|---|---|---|
| 0-3 days | 91 | Just been in |
| 4-14 days | 65 | 7 and 14-day alerts watch here |
| 15-30 days | 30 | 14-day escalation zone |
| 30+ days | 36 | Pre-cancellation territory |
The system runs on Coach Mike's existing TeamUp account
TeamUp is where his bookings, attendance, and memberships live. Personify pulls that data, every hour, into a read-only pipeline. The AI sees attendance the moment TeamUp does, and that single feed powers eight of the twelve workflows.
TeamUp, read-only
Bookings, attendance, memberships
- Attendance records
- 21,400
- Class events
- 2,540
- Member records
- 700
- Membership history
- 2,540
Engagement system, live
Listens, decides, drafts
- Syncs per day
- ~49
- Database tables
- 47
- Templates ready
- 237
- Workflows active
- 12
The right message, coach-approved
In Coach Mike's voice, on time
- Low-attendance return
- 90%
- 7-day return rate
- 65%
- Sent without edits
- 97%
- Hours back per week
- 13-19
Each TeamUp signal triggers a specific workflow
| TeamUp signal | Workflow | Result |
|---|---|---|
| Last attendance > 7 days | 7-Day Absence Alert | 65% return |
| Last attendance > 14 days | 14-Day Escalation | 27% return |
| Attendance frequency declining | Low-Attendance Warning | 90% return |
| Status: no_show same-day | No-Show Follow-Up | Same-day touch |
| Booked but not yet attended | Booking Reminder | Fewer empty slots |
| Sessions this week count | Weekly Sunday Check-In | Tone adapts |
| Induction class booked + attended | Induction Journey advances | 12 stages |
| Member joined 6 weeks ago | Referral Ask | Right-moment timing |
The catch rate
Three workflows handle the silent-churn pipeline. Each fires automatically, drafts in Coach Mike's voice, and waits for him to approve.
Attendance warning
Caught before the drop
Friday 4pm, the system flags members whose frequency is sliding.
returned within two weeks (265 of 295 caught)
7-day alert
Caught at one week
When attendance passes 7 days, the system drafts a warm nudge.
returned within two weeks (88 of 135 caught)
14-day escalation
Caught at two weeks
Past a fortnight, the tone shifts. More direct. Offers a chat or rebook.
still returned (36 of 134 brought back)
His voice, at scale
The system drafts. Coach Mike approves. He doesn't write.
The system drafted 2,315 personalised messages last month. Coach Mike sent 2,236 of them without changing a word. On the 79 he did edit, the pattern is consistent: he didn't rewrite, he trimmed.
AI first draft
characters
Final sent by Mike
characters
Average cut: 206 characters. The structure stays, the voice gets sharper. That's the human-in-the-loop story, quantified.
Approval rate by message type
Templated workflows
Zero edits, held all-time
Mass campaigns
Every one sent untouched
Check-in responses
864 of 864 as drafted
Free-form replies
760 of 839 on hardest case
13-19 hours of coach time back, every week.
Each message used to take 4 to 6 minutes. Not just typing. Pulling up the member's check-in data, scanning their intake forms, researching the answer, then writing it in his voice. The system does the context-gathering and the first draft. Mike approves.
The build
Twelve workflows on cron. A twelve-stage onboarding journey. 237 message templates, 209 of them Meta-approved. One human approval gate. Every outbound message goes through Coach Mike before it reaches a member.
Twelve workflows, always on
7-Day Absence Alert
14-Day Absence Escalation
Low Attendance Warning
Weekly Sunday Check-In
No-Show Follow-Up
No-Response Chaser
Booking Reminder
New Member Journey
Induction Journey
Bimonthly Progress Review
Referral Ask at 6 Weeks
Google Review Request
Human approval on every message. The system drafts. Coach Mike approves, edits, or rejects. The AI never speaks alone.
Twelve-stage onboarding journey
Every new member runs through the same 4-week sequence. Welcome, first-session follow-up, goal pack, weekly check-ins, a booked progress call. By week four, they're in the same rhythm as the rest of the gym.
Week 1
Welcome + first session + goal pack
Week 2
Weekly check-in + booking prompt
Week 3
Check-in + progress call booked
Week 4
Final check-in + graduation
More Personify case studies
Other coaches and consultants who scaled with Personify.
Career coaching
Lucy Gilmour
1,056 conversations in 9 days. $8,800 in 24 hours. Course price raised from $39 to $199.
Read the case studyEndurance coaching
Janis
840+ conversations per month across 10 countries and 7 timezones. Sub-30-second response time.
Read the case studyWant a system like this for your business?
Book a 30-minute call. We'll map out what your AI version looks like and what it would take to build.