Most teams try to build a Version One. You don’t need a Version One. You need an answer.
I. The High Cost of Guessing
Every product is a chain of assumptions: who the customer is, what hurts, why your fix is better, how they’ll find it, and why they’ll pay.
If any link snaps, the rest doesn’t matter.
Big companies miss, too. The first iPhone shipped without an App Store. It became inevitable after they found the missing piece. Point is: nobody sees the right combination up front. You discover it.
Burning months on the wrong thing is expensive. Burning your team’s trust is worse. The only rational goal is simple: spend less time being wrong so you have time left to be right.
II. MVPs Are Experiments, Not Early Versions
Somewhere along the way “MVP” got translated as “ship a weak first version.”
That’s not the job. The job is to test the core bet with the least amount of time, code, and ego.
Race teams don’t build an entire car to see if a new engine works.
They put the engine on a stand and run it hot. If it holds, then they bolt it into something pretty.
Your MVP is the stand.
Design the MVP like a lab test, not a launch:
- Hypothesis: “Underbanked patients will pay cash at retail if given a scannable bill.”
- Signal: “≥10% of targeted bills are paid via the retail option within 14 days.”
- Minimum setup: “Add a QR to 200 bills; route payments to a basic ledger; manual ops allowed.”
- Timebox: “Two weeks, then decide.”
- Decision rule: “If signal < 5%, stop. 5–10%, refine. >10%, build.”
That healthcare example? An intern pasted QR codes onto a handful of bills. Cost: about $100.
They got an answer in days, not quarters. That’s an MVP — not perfect, but useful.
Ignore the alphabet soup (MMP, MLP, MAP). If the test gives you confident signal on the risk that can kill you, it’s viable.
If it doesn’t, it’s theater.
MVP = the smallest thing that can prove you’re not kidding yourself.
Rough edges are fine. If you truly deliver 5–10× the value, customers forgive rough edges.
What they won’t forgive is wasted time.
III. Pivoting: When the Evidence Says You’re Wrong
You need two conflicting muscles: stubbornness about the problem and detachment from your solution.
Fall in love with the problem; stay flexible with the fix.
Sunk cost will whisper in your ear: “We’ve already built so much.”
Yes — and that’s the tuition you paid to learn what doesn’t work.
A simple pivot test:
- Did we test the riskiest assumption?
- Was the test clean? (Right users, clear threshold, enough data.)
- Did we hit the threshold?
- Yes: Double down on what worked.
- No: Change one major variable (who, problem, channel, pricing) — not five.
Pivoting isn’t flailing. It’s adjusting to what the evidence actually says.
Remember the iPhone: the hardware wasn’t enough; the App Store unlocked the system.
The win was in the combination, not the parts.
IV. PMF: Evidence, Not Vibes
Product-Market Fit isn’t a feeling. It’s the market pulling the product out of your hands faster than you can push it.
How PMF tends to look (depending on your model):
- People come back on their own. Retention curves flatten above zero.
- Would-miss score: ≥40% say they’d be “very disappointed” without it.
- Revenue says yes: People pay, keep paying, and ideally pay more.
- Sales motion gets easier: Inbound rises, sales cycle shortens, excuses shrink.
- Unit economics stop gaslighting you: LTV:CAC ≥ 3:1 with believable payback; NRR > 100% in B2B.
- Support noise changes: Fewer “what is this?” tickets; more “can it also do X?”
If you don’t see these patterns, you don’t have PMF yet. That’s not failure — it’s feedback.
Run the next experiment.
The best proof of value is a receipt. A nice side effect is cash.
V. Field Guide: Run the Loop Like an Adult
1) Write the bet.
Customer, problem, value delta (“5× faster than Excel”), and the moment of proof.
2) Choose the smallest test that can fail.
If it can’t fail, it can’t teach you anything.
3) Timebox decisions.
Two-week sprints to answers, not to burndown charts.
4) Change one thing at a time.
If everything moves, nothing learns.
5) Log the scars.
Keep a one-page “lab notebook”: hypothesis, setup, result, decision. Future-you will thank you.
VI. Conclusion: Evidence Over Ego
Stop trying to build a perfect Version One.
Build a sequence of honest experiments that make the final product obvious.
If the data says pivot, pivot.
If it says double down, get boring and repeat what worked.
If it says “no one cares,” stop romanticizing the idea and change the idea.
Do the smallest thing that gives you a real answer. Then do it again next week.
Give me a call when you’re tired of building the wrong thing, and we'll get through it together.