From Idea to MVP to Funded: A Founder's Honest Playbook
The most expensive sentence in any startup is "let's just build it and see." I've said it. I've watched friends say it. It is the sound of a founder falling in love with their solution before they have any evidence the problem is worth solving.
I've taken several companies from idea to MVP to market, and sat in accelerator and fundraising rooms on both sides of the table. The single pattern I'd undo if I could is the order I worked in. Most founders run it as idea, build, launch, then go looking for customers. Every credible playbook — Steve Blank's customer development, Eric Ries's Lean Startup, Y Combinator's own library — inverts that. It's problem, then customers, then evidence, then build. The MVP comes after customer discovery, not before it. Get that sequence wrong and everything downstream is just expensive motion.
CB Insights looked at why startups die and found that 42% cited "no market need" as a top reason — the single most common root cause. That number is not about bad engineering. It's about building something nobody was waiting for. So before you write a line of code, earn the right to.
Fall in love with the problem, not your solution
YC's most-repeated lesson is brutally simple: founders hold too tightly to their first solution and too loosely to the problem. The market is the durable asset. Your v1 product is disposable, and it is almost always wrong. You don't discover the right product in a planning doc — you discover it by launching something, talking to real users, and iterating toward what they actually pull on.
The best ideas, as YC frames it, share three traits: it's something the founders themselves want, that they can build, and that few others realise is worth doing. Notice what's missing — a polished pitch. Look for problems you have organic insight into, not made-up ones you reverse-engineered from a trend.
Steve Blank's line is the one I tattooed on my forehead, figuratively: "There are no facts inside your building." A startup is a temporary organisation searching for a repeatable, scalable business model — not a smaller version of a big company. So treat your assumptions like a scientist treats a hypothesis. Write down what you believe to be true about the customer, the problem, and what they'll pay. Then get out of the building and go disprove it. In KL, that means coffee in Bangsar with ten real prospects, not a survey blast to your WhatsApp contacts who'll be too polite to tell you no.
The MVP is a learning instrument, not a small product
Here's where I burned the most money early on: I thought the MVP was the cheapest possible version of the product. It isn't. Eric Ries is precise about this — the MVP is whatever yields the maximum validated learning for the minimum effort. It's the fastest possible trip through the Build-Measure-Learn loop. It's an experiment that answers two questions: should this be built at all, and can we build a sustainable business around it?
That reframing changes what you build. If a landing page and ten manual onboarding calls answer your riskiest question, that's your MVP — not three months of engineering. Scope ruthlessly. For everything someone asks you to add, the default answer is no until a real user's behaviour forces a yes. Build the one thing that tests your core hypothesis. Refuse the dashboard, the settings page, the second integration, the "nice to have" everyone nods at.
And measure with actionable metrics, not vanity ones. Sign-up counts feel good and tell you nothing. Whether the people who signed up came back, and whether your business-model drivers actually move, is the signal. When those metrics aren't moving no matter how hard you push, that's not a reason to push harder — it's the signal to pivot, to make a structural change and test a new fundamental hypothesis.
Do things that don't scale — on purpose
Paul Graham's essay on this should be required reading, and I'd add: the unscalable work is not a phase to rush through, it's a learning advantage and arguably part of your moat. Startups don't take off on their own. You crank the engine by hand. You recruit your first users one at a time, and you delight a few of them so intensely they can't shut up about you.
The named patterns are instructive. The "Collison install" — Stripe's founders setting up the product on a customer's device themselves, on the spot, rather than emailing a link. Airbnb's founders knocking on doors and photographing listings. These weren't growth hacks. They were how the founders learned what to build. Graham's reassurance is one I lean on: "I have never once seen a startup lured down a blind alley by trying too hard to make their initial users happy."
AI lowering the cost of shipping makes this more true, not less. When everyone can stand up a working product in a weekend, the differentiator is no longer the build — it's the founders who are still in the room with the customer. Start in a deliberately narrow market, what Graham calls a "contained fire." Own a small group completely before you reach for the big one.
Product-market fit is a number, not a feeling
"We feel like we've got traction" is not an acceptable answer, and most founders fake PMF to themselves. Polite interest looks a lot like fit until you try to scale on it. Then it evaporates, and you've already hired the sales team.
Make it measurable. The Sean Ellis test: survey users who've used the product at least twice recently and ask, "How would you feel if you could no longer use this?" If 40% or more say "very disappointed," you likely have PMF. Slack reportedly scored around 51% at roughly half a million paying users. It's directionally useful from about 40 responses, so you can run it early. Rahul Vohra's team at Superhuman built a whole repeatable engine around that one question — segmenting the high-expectation users, doubling down on what they loved, systematically converting the merely "somewhat disappointed."
Triangulate it. Pair the survey with a retention curve that flattens rather than decaying to zero, and an NRR benchmark — for bottom-up SaaS, around 100% is good and 120% is great; for enterprise, 110% good, 130% great. PMF, as Lenny Rachitsky puts it, shows up as sudden pull, compounding pull, or a milestone that proves the model. If you're squinting to see it, you don't have it yet. Treat that 40% as a gate that must be passed before you hire sales, raise a big round, or spend on paid acquisition.
Premature scaling is the disease; running out of cash is the symptom
This is the reframe I most want ASEAN founders to internalise, because the region is a little obsessed with raising rounds. Yes, 70% of failed startups cite running out of cash. But CB Insights is clear that's the final symptom, not the root cause. The Startup Genome research names the actual disease: premature scaling — spending and headcount running ahead of validated demand. In their data it accounts for the majority of high-growth startup failures: the great majority of companies that scale prematurely never cross US$100,000 a month in revenue, while properly-scaled startups grow many times faster.
Read that again. Raising too much, too early, before PMF, doesn't accelerate success. It accelerates death. The big round buys you a bigger team, a bigger burn, and a faster clock — all aimed at a product you haven't validated. The discipline is counterintuitive but it's the whole game: don't pour fuel on the fire until you're sure the fire is the right one.
Fundraising is a story backed by traction, run as a tight process
Investors don't fund decks. They fund a story the partner who likes you can repeat, accurately, in the partner meeting you're not in. Traction is the backbone of that story. First Round's framing — drawn from a portfolio that has reportedly raised many billions in follow-on capital — is to shift from "we win customers" to "we are the inevitable winner of a large market that is unfolding right now." That maps directly onto Sequoia's "Why Now" slide, the one great companies nail: the historical shift that makes this idea possible today and not five years ago.
Run the raise as a tight, time-boxed process, not an open-ended shop. First Round's tactics are the ones I'd actually use. Batch your outreach — around five investors at a time, mixing top targets with lower-priority firms, so meetings cluster and momentum becomes visible. Remember the 10/90 rule: only the top 10% of rounds get competitive auctions; for the rest, your job is to get one or two partners genuinely, personally excited. Lead with traction before the midpoint of the deck. Drop the competitor grid in favour of differentiation framed from the customer's point of view. And name your own weaknesses first — admit the unsolved channel or the CAC you haven't cracked. Naming it builds more credibility than any spin, and it pre-empts the burning questions they were going to ask anyway.
A word on grants, for Malaysian and ASEAN founders
We have something many ecosystems envy: real non-dilutive money. Cradle has funded over a thousand Malaysian tech companies, with CIP grants designed to take a startup from MVP through to commercialisation; MDEC runs digital and AI grants that can cover a large share of project cost; and recent national budgets have continued to back digital-acceleration funding. (Programme names, ceilings and eligibility change often — confirm the current terms before you bank on a figure.) Used well, this money de-risks the MVP without giving away your cap table.
But grants are a double-edged sword, and I've watched good founders get cut by them. Grant-chasing quietly optimises for application criteria and milestone reports instead of paying customers. You end up building the product the grant wanted, not the one the market wants. Take the money — then keep your actual scoreboard fixed on customers and revenue. The grant funds the experiment. It is not the experiment's result.
Where to start
If you take one thing from a founder who's made most of these mistakes: invert the order. Spend the next two weeks not building. Write down your riskiest assumption about the problem, get out of the building, and go test it with ten real people. Let what you hear shape the smallest possible MVP that answers "should this exist?" Measure fit as a number before you scale anything. And when you raise, raise on traction and a why-now story, run tight, and name your own weak spots first.
The two most expensive founder mistakes — building before validating, and scaling before fit — share a root cause: nobody stayed in the Build-Measure-Learn loop long enough to learn. That's the gap we built Appcellen to close. We design, build and run platforms — we don't ship a spec and disappear — which means we stay in the loop alongside founders, validating and iterating rather than handing over and leaving. If you're somewhere on the road from idea to raise and want a partner who's walked it, we'd be glad to talk.
