Last time I shipped into silence. This time I'm selling before building.
Three weeks ago I posted here about JobHunting, my interview-prep tool that shipped into total silence. That thread changed how I work — I stopped announcing and started being present where the pain actually lives.
This post is about the other product, and the opposite experiment.
My dog turned five this year. His medical history lives in four places: feeding notes in a chat thread, vaccine dates in photos, the last blood test in a drawer, symptom changes in my memory. Every vet visit starts with "when did this start?" and I never know.
PLANET is the thing I want instead: one shared timeline for a pet's whole life, for households where more than one person does the care. It answers two questions: "who did what today" (so meds and walks don't get missed or doubled), and "what's the full story at the vet" (a summary you hand over in one tap instead of scrolling your phone in the waiting room).
This time I took the thread's advice before building:
-
Free tools first, no signup — a symptom checker (20 common symptoms, when to call the vet), a pet card maker, and a shared vaccine calendar. The "build the thing that answers what your buyers already type" play.
-
A real checkout before the real app. 100 founding seats, first tier S$29.99 one-time, price rises as each tier fills. Live counter on the site, no fake countdown. Current count: 0. That's the honest part — I'd rather know now than six months from now.
-
A free pilot for 10 families running alongside, no payment during pilot. Because the real risk isn't "will anyone pay" — it's whether a second caregiver actually joins and sticks for two weeks. If the collaboration doesn't happen, payments don't matter.
And the mistakes-already-made list, since that worked so well last time:
- First checkout leaked my backend URL to users. Rebuilt same-origin.
- The Lemon Squeezy webhook payload had a custom value in the wrong nesting — it silently failed while looking fine on the dashboard.
- The 100-seat cap started in code until I saw the race condition; it lives in a Postgres transaction now.
What would genuinely help:
- If someone else shares the care of your pet — what's your current "system"? Chat threads count. Mine was one.
- IH people: the site shows "0 of 100 founding seats paid" next to a live checkout. Honest or dead on arrival? I keep going back and forth on whether to show the zero.
Free tools + both paths: https://www.joinplanet.pet
The strongest part is that you’re testing two separate risks before scaling: willingness to pay and whether shared caregiving actually happens. That makes the validation unusually concrete.
This is iteration speed reduced to its core. Building in silence = you have to wait for release to measure if anyone cares. That's a weeks or months measurement latency. Selling before building = you measure market interest before you commit code.
The real win isn't that you save build time (you do). It's that you collapsed your measurement latency from "weeks after launch" to "before you start." You went from false momentum to real signal immediately.
Your measurement latency determines your actual iteration speed. Building fast doesn't help if you can't know whether you're building the right thing until 3 months later. Selling before building is just measurement-first thinking - you're getting signal before the big bet.
This unlocks the ability to actually iterate fast. Not fake-fast (shipping things quick) but real-fast (knowing what to build next because you already measured what works). Once you know what works, building it is the easy part.
'False momentum' is the exact phrase for what the first launch felt like — daily commits, a shipping cadence, and zero information. I was optimizing a loop that never closed.
One honest catch on the fast signal though: at small n it's noisy. A checkout with 0 sales after a week can mean "nobody wants this" or "the twenty-five people who'd want it never saw the page." The measurement is only fast once the traffic side exists too — which is why the pilot and the founding seats run in parallel. The seats measure willingness to pay; the 10 families measure whether the collaboration actually happens when a second caregiver is invited. Different question, same measurement-first logic.
And credit where it's due: the discipline came from your question on the last thread — "did they interact with the core insight" instead of "did they sign up." Payment is just the loudest version of that question. I'll post the pilot numbers here as they come in, ugly or not.