A flight plan.
This is the document that comes out of stage 02 — the thing you actually get before anyone writes code. It says what the first launch is, what it is not, and what we will know at the end of it.
The mission below is fictional. No client work is published here without permission, and none is published yet.
The mission
Independent physiotherapists in South Africa run their practices on WhatsApp, a paper diary and manual EFT reconciliation. Booking a session takes four messages. Roughly one appointment in six is a no-show, and chasing payment afterwards is the part everyone hates.
The hypothesis: a solo practitioner will pay a monthly fee for a booking link that takes card payment up front, if it removes the back-and-forth and the no-shows.
The riskiest assumption: not whether practitioners want this — they say they do — but whether their patients will pay up front for a service they are used to paying for afterwards. If patients abandon the booking at the payment step, the whole model changes.
What counts as a yes: across 12 pilot practices in 6 weeks, at least 60% of started bookings reach payment, and at least half the practices are still using it in week 6 without being nudged.
The smallest product that tests it
One workflow, end to end: a patient books and pays; the practitioner sees it. Everything else waits.
In the first launch
- Practitioner sign-up and a public booking page
- Availability set as weekly recurring hours
- Patient books a slot and pays by card (Yoco)
- Confirmation and reminder by WhatsApp
- Practitioner’s day view, with cancel and refund
- Funnel analytics on every step of the booking
Deliberately not yet
- Multi-practitioner practices. 80% of the market is solo. Adds roles and permissions to every screen.
- Calendar sync. Two weeks of work and a permanent support burden; the day view answers the same need for now.
- Clinical notes. Regulated, and irrelevant to the hypothesis.
- Medical-aid claims. The real long-term prize, and the fastest way to spend six months not launching.
- A native app. Patients arrive from a WhatsApp link. A web page is the correct shape.
- An admin panel. Twelve pilot practices can be managed from the database by us.
Systems & decisions
Boring, cheap, and easy to hand to somebody else. Every choice here is reversible except the first one.
Milestones
Open questions and risks
Patients may not pay up front
The central risk, and the reason the funnel analytics are not optional. If drop-off at payment is severe, the next mission is a deposit rather than full payment — a small change we have deliberately left room for.
WhatsApp template approval is outside our control
Meta approval has taken between two days and three weeks in our experience. Submitted in week 1; email is the fallback so the launch date does not depend on it.
Twelve pilots is a small sample
Enough to kill the idea, not enough to prove it. A clear yes here means a second, larger mission — not a scale-up.
Refund policy is a business decision, not ours
Needed before launch. Flagged in week 2 so it does not become a week 8 emergency.
What we will know on 14 May
Whether patients will pay before a session. Whether practitioners keep using it once the novelty wears off. Which of the six deferred features gets asked for first — which is the cheapest possible way to find out what the second mission should be.
If the answer is no, it will have cost nine weeks and one clear paragraph will explain why. That is the point of launching early: reality gets a vote.
Kestrel is a fictional mission written to show the shape of the document. Every figure in it is invented.
Bring us the mission.
Tell us the idea, the problem or the half-built thing. We’ll tell you what the smallest worthwhile first launch looks like.