How to Start: from Idea to First Customer

The first 30 days are about validation, not building.

Most founders skip validation because building feels productive. You can see a feature ship. You can deploy something. Validation, by contrast, feels slow and slightly humiliating — you have to admit you don't know what you're doing yet. So founders skip it and go build.

This is the most expensive activity in startup-land. Building before validating is like writing a novel before you know who reads it. The first month is over before you know whether the premise works.

Validation is a different mode of work. It's mostly conversations, plus a landing page, plus the discipline to keep listening when the answer you're getting doesn't match the answer you wanted. The output isn't code; it's a sentence you can defend, like: "I'm building X for Y because they currently do Z and hate it." If you can't say that sentence, you're not ready to build.

Here's the bar. If you can't get 10 strangers to nod when you describe the problem, you don't have a problem worth solving yet. Not 10 friends. Not 10 LinkedIn connections you already know like your work. Strangers. People who have no incentive to be polite. The nodding has to be involuntary — the moment of "oh yeah, that's annoying" — not the polite "interesting, tell me more" nod.

The cost of skipping this is not zero. It's the next six months of your life. Every founder I've worked with who skipped validation came back at month four asking why nobody is buying. The honest answer is always: because the problem you imagined was different from the problem they have. You don't fix that with marketing. You fix it by going back to month one and doing the thing you skipped.

Pick a market you can talk to in 5 minutes.

When founders describe their target customer to me, I time them. If it takes more than 5 minutes to explain who buys, the market is too vague. Too vague means you can't write a cold email. You can't write an ad headline. You can't even write a tweet that lands.

Niche down until your customer is unambiguous. "AI tools for accountants" beats "AI tools." "AI bookkeeping for solo accountants in the US who use QuickBooks" beats "AI tools for accountants." Each layer of specificity makes everything downstream easier — the messaging writes itself, the channels become obvious, the price point becomes defensible.

Founders resist this because it feels like they're shrinking their market. They're not. They're shrinking their initial market. A specific persona is a wedge — once you've nailed the workflow for solo QuickBooks accountants, the same product expands naturally to small firms, then mid-sized firms, then bookkeepers, then the broader finance ops world. But you can't expand from nothing. Pick the wedge.

The 5-minute test has a second use: it tells you whether you actually understand the customer. If you can't describe their day in concrete detail — what software they open in the morning, who interrupts them, what makes them sigh — you don't know enough yet. That's a gap to close before you write a line of code, not after.

A specific persona beats a category every time. A category is a slide in a pitch deck. A persona is a person you can call.

The minimum viable validation: 10 conversations.

Here's the bar I hold founders to before they're allowed to build anything. 10 conversations. 5 of them must be paying customers of an existing alternative. 3 must say they'd pre-pay before the product exists. 2 must offer to introduce you to 5 more people. If those numbers don't show up, the idea isn't validated yet — kill it or pivot.

Each constraint is doing real work.

Paying customers of an alternative matters because it's the only proof the problem is worth money. Free users will tell you anything. People who pay $50/month for a worse alternative are showing you they have a budget — and that budget is yours to win.

Pre-pay before the product exists filters the polite from the committed. Anyone can say "yeah, I'd buy that." Almost nobody will hand you a credit card for vapor. The 3 people who do are the most valuable signal you'll get this year. They're also your design partners — give them an embarrassing discount in exchange for input, and ship for them first.

Two intros each, totaling 10 more leads is how validation compounds. If your first 10 won't introduce you to anyone, you don't have evangelists. No evangelists, no growth loop, no business — at least not without burning ad money to substitute for word-of-mouth.

Run this gate strictly. Most ideas die at this stage, and that's the point. The 10 conversations are how you discover the idea was wrong cheaply, not the validation theater you do to confirm an idea you already love.

Naming, branding, and the legal stuff: do it later.

Most first-time founders waste 2 weeks on this. They spend a Saturday on Squadhelp. They read about LLC vs. C-Corp on Reddit. They redo the logo three times. None of it produces a paying customer. None of it is the work.

The right time to incorporate is the day before you take money. Not a week before. Not when you "have an idea." A day before — because that's when there's actual money to receive. Until then, you're a person on the internet, and that's fine. The Stripe account, the bank account, the Delaware C-Corp, the lawyer, the cap table — all of that exists to handle money. No money, no need for the apparatus.

Pick a name that's available and not embarrassing. That's the bar. The .com is preferable but a .ai/.co is fine for v1. Domains are cheap; you can rename later if you have to. The companies you've heard of mostly have boring names — Zoom, Slack, Stripe, Notion. They became iconic because of the product, not because the founder spent three weekends on naming.

Same logic for branding. A clean Helvetica wordmark on a white background is fine for the first 6 months. A real design system can wait until you have signal that your customers care what you look like. They don't, until they do, and you'll know when.

Move on. The thing that determines whether your company exists in 12 months is whether you talked to enough customers, not whether you incorporated in Delaware or Wyoming.

When to start writing code.

Only after the previous four sections are done. The 10 conversations are complete. The persona is unambiguous. You've validated, niched, and ignored the legal/branding rabbit hole. Now you can build — but not the product.

Build a landing page that promises the product. One page. Hero, value prop, three bullets, an email signup. Use Framer or a Next.js template; don't custom-design it. The whole thing should take a day, not a week. The page describes what your customers will get; it doesn't actually deliver it yet. That's fine. You're measuring intent, not building a business yet.

Run cheap traffic to it. $50 of Meta ads pointed at your persona. A few cold emails to the 10 people from your validation interviews, asking them to share. A LinkedIn post. Whatever channels match your audience. The metric is email signup rate per visitor. If 5%+ of qualified visitors enter their email, you have signal. If it's under 2%, your message isn't landing — fix the page before you write more code.

This is how I work, every night, on every company. Page first. Signups first. Then build the smallest version that fulfills the promise on the page — and ship it to the email list as an early-access waitlist.

The founders who do this in order get to product-market fit fastest. The ones who skip ahead and build first spend month four staring at an empty signup form, wondering where everyone is. The order matters more than the speed.