One diagnostic, in full
They were paying for leads nobody followed up.
A music school in India, running ads, taking enquiries steadily, stuck at about half its revenue target. The founder’s conclusion was that the top of the funnel needed to be bigger, and next quarter’s budget was about to go into more advertising.
177
Leads paid for, contacted once, then dropped
0%
Show-up rate nobody had calculated
0.8%
Enquiry to enrolment at the time
Every figure comes from the client’s own records. Client name withheld.
The diagnostic
What the records said.
Before recommending anything we asked for the numbers they already had: the enquiry log, the fee ledger, the salary sheet and the student roster. Four spreadsheets that had never been read against each other.
Enquiries
259
Called back
167
Visits booked
20
Visits attended
0
Enrolled
2
Twenty visits were booked and nobody arrived. The two enrolments came without a visit. The open pipeline broke down as 69 marked follow-up, 67 no answer, 28 visit to be scheduled and 13 visit scheduled. The oldest untouched lead was five months old, studio occupancy sat at 15% against fixed rent, and ten students were on break, paying nothing, with nobody chasing them.
They did not have a demand problem. They had a conversion, retention and utilisation problem, and all three were automatable.
The findings
Three leaks, all automatable.
Leak one
The leads they had already bought
Instant reply the moment an enquiry lands, multi-touch nurture so nothing goes cold, reminders before a booked visit, and automated re-engagement of no-answers and no-shows. Converting 5 to 10% of the 177 idle leads is 9 to 18 students, or +12% to +23% on revenue, from leads already paid for.
Leak two
An empty studio against fixed rent
Studio capacity is 132. Twenty studio students were active, 70 across studio and online, and studio margin sat at 34%. Teaching capacity already existed: one instructor carried 606 lessons while two others carried 104 and 8. Growing 20 to 40 students, still only 30% full, adds +26% at close to pure margin.
Leak three
Churn nobody was watching
Ten students on break means 15% of revenue paused, and in this sector break is how a school loses a student. Flagging attendance and payment slips before they leave, plus automated win-back, recovers roughly +7% and surfaces losses that were previously invisible until they had already happened.
Modelled at engagement, not a result
What the three levers were worth.
This is the model we put in the readout. It is a projection from their own records at their actual fee per student, not a benchmark and not an outcome we have measured.
| Lever | Assumption | Modelled |
|---|---|---|
| Measured baseline | At engagement | 100% |
| Lead conversion | 5% of the 177 idle leads, 9 students | +12% |
| Studio fill | 20 to 40 students, still 30% occupancy | +26% |
| Reactivation | Half the students on break | +7% |
| Realistic, about 90 days | Conservative assumptions throughout | +45% |
Alongside the revenue, 40 to 60 hours a month returned, thirteen of them from reconciling attendance across spreadsheets by hand.
Delivered
What got built.
The problem
Seventy students on WhatsApp
The school ran on WhatsApp threads and the owner’s memory.
Built
A second brain and one dashboard
A second brain of the school’s context, one dashboard for schedules, attendance and fees, and separate portals for teachers and students, with the lead follow-up and win-back modules from the diagnostic built in.
Designed for
The whole school in one place
One view for the owner of schedules, attendance, fees and leads, and teachers managing their own classes instead of messaging the owner.
The next case
Two founding slots at 50 to 300 people.
The case above is a small business. The first two diagnostics at companies of 50 to 300 people run at $2,500 instead of $5,000, in exchange for a named case study like this one, with real numbers, a short video testimonial and two introductions. Agreed in writing before we start.
Governance
How AI is allowed to behave in your company.
The governance is built before the capability, because a system that does something nobody sanctioned loses the room once and never gets it back.
Rule 01
Nothing fires unreviewed
Every outbound action lands in an approval queue first. The agent proposes, and a person releases.
Rule 02
Autonomy is earned, not granted
An action type runs unattended only after 30 days under a 5% override rate, and it is demoted after a single serious miss.
Next step
Find out what yours would show.
This is the diagnostic in full, from the client’s own records. Twenty minutes to find out whether the same method is worth running in your company.
$5,000 diagnostic, credited in full · you keep the roadmap either way · you own everything we build