Booking was neverthe problem.
Slate was going to be a booking page for independent physios. Ten sessions said the calendar was the easy part, and the thing people wanted was a reply. Here is the whole trail, including what got killed.
The idea, in one line
A booking tool for independent physios who currently run everything through WhatsApp.
The physio who described it had around sixty regulars, a phone that never stopped, and a Sunday evening spent copying appointments into a paper diary. She wanted what the big clinics have: a booking page, a calendar, confirmations that send themselves.
That is a clear brief. It is also a guess about which part hurts.
One thing worth saying up front: Slate is one of my own nine apps, not a client engagement. The regulars were already reachable, so the sessions ran the same week the build finished. On someone else's idea that is the part you cannot count on, which is why testing sits outside the two weeks and can be added on.
People would rather book themselves than ask.
The exciting part was the calendar. The risky part was six words further down the brief: "so she stops answering messages." Everything else depended on sixty regulars choosing a form over a thread, and nobody had ever asked them to.
Three things were ruled out as not the risk, on purpose:
- Taking payment. Everyone already paid in the room. Adding cards would have tested a problem nobody had.
- More than one practitioner. A real constraint one day, an excuse to build a calendar system now.
- Insurance codes. Fiddly, visible, and irrelevant to whether anyone books at all.
Agreed before any testing
Three sentences, written down before there was anything to look at, so the bar could not move later.
Go
Six of ten regulars book their next appointment through the page unaided, and nobody messages afterwards to check it worked.
Adjust
They book, then message anyway. The booking works and something else is missing.
Stop
They open the page, close it, and go back to WhatsApp. Booking was never the friction.
One flow, made real
See this week, pick a time, done. Usable on a phone, in ninety seconds, by someone standing in a hallway.
What got left out on purpose:
- Accounts and passwords. A first name and a phone number identify a regular well enough for two weeks.
- The practice-side calendar. A spreadsheet did that job behind the scenes. Nobody was testing the spreadsheet.
- Cancellation rules, reminders, profiles, a policy page. All real one day. None of them the question.
What ten sessions showed
Nine out of ten cleared the bar the tool was built to clear. Then seven of them opened WhatsApp and typed the thing the tool had just made unnecessary.
"I've booked it, but can you just confirm you got it?"
"I don't mind messaging her. I mind not knowing if she's seen it."
Booking was never the friction. Waiting was. The confirmation screen said the appointment existed. It did not say a person knew about it, and that is the reassurance people were actually chasing.
The call: adjust
The rule said go only if nobody messaged afterwards. Seven did. So this is an adjust, and the flat version of that is worth more than a generous one.
Adjust
The problem is real and the shape is wrong. This is not a booking system with reminders bolted on. It is a confirmation system that happens to take bookings.
The change is one sentence: lead with the reply, not the calendar. A regular should get a message from a person, in the thread they already use, and the booking page should be the quiet mechanism underneath it rather than the thing being sold. The same two weeks can run again on that version, against a new bar.
What I would change and kill
What the two weeks bought
A named change instead of a hunch, for the price of ten conversations and one flow. The version that would have shipped without this was a well-built calendar for a problem that was never about calendars.
What I would change if this were a paid product
Make the confirmation the product. The reply lands in WhatsApp, from the practice, within seconds. The booking page becomes the mechanism underneath it instead of the thing on the poster.
What I would kill
The week view. It took four days and two of the ten people used it. Everyone else took the first sensible slot they were offered. It survived that long because it looks like the thing a booking tool is supposed to have, which is the worst reason for anything to exist.
Where the two weeks went
Days one and two
Find the risky part
The brief said calendar. The questions said reassurance. The decision rule got written before there was anything to look at.
Days three to eight
Build one flow
Three screens, real times, no accounts. Four of those days went into a week view that later got cut, which is the tuition.
Days nine to twelve
Ten sessions, then the call
Ten regulars, one flow, no coaching. Fast only because the list was already mine. The call written against the rule, with the evidence underneath it.
Got an idea with a risky part?
Ten minutes of questions about the risky part, then a straight read on whether this fits. The answers are yours to keep either way.
Tell me about your ideaHere it is. It’s yours either way.
Nothing has been sent yet — this site has no server to send it to, and a form that says “sent” when it means “discarded” is worse than no form. Copy it or open it in your mail app, and it reaches me the moment you hit send. Keep a copy either way: the answers are more use to you than they are to me.