The first website for a business that does not exist yet

A founder we've watched do this twice now spent six weeks on her pre-launch site both times — a designer, three rounds of copy, a logo exploration that ran to eleven options. The first time, the site launched two days before she changed the pricing model entirely, from a subscription to a one-time fee, so the site's structure — three tiers, a comparison table, a "cancel anytime" line in the footer — described a business that no longer existed. The second time she knew better and still took five weeks, because "better" meant a nicer design system, not a smaller decision.

This isn't a story about a founder who didn't know what she was doing. It's the standard failure mode for anyone building before the first sale, and it survives contact with experience because the mistake doesn't feel like a mistake while it's happening — it feels like doing the work properly.

The site is not a milestone, it's an instrument

Give the counter-argument its due first, because it's not wrong on its own terms: a considered site does make a real first impression, and for a company that already has customers, that impression is worth the weeks it costs. We've made that case at length — see what a company website is actually for, which walks through the four jobs a business site can do and how to name yours before you spend anything.

But all four of those jobs assume the business is the fixed part and the site is built to serve it. Before the first customer, that's backwards. The business is the thing under test. The offer, the price, the person it's for — all three are guesses, however informed, and the site's only honest job is to expose the guess to someone who can prove it wrong. A site built to look permanent while the thing it describes is still a hypothesis is decorating a number you haven't validated yet. Every hour past the minimum needed to state the offer is an hour spent making the wrong guess look more finished.

The page has to be specific enough to refuse

Here's the test that actually separates a working pre-launch page from a decorative one: could a stranger read it and say no for a real reason?

"We help small businesses grow with data-driven marketing solutions" cannot be refused, because there's nothing in it to refuse. Nobody reads that sentence and thinks "not for me" — they think "I don't know what this is," a different and much less useful reaction. It generates polite curiosity and zero information.

Compare: "A monthly report that tells a five-person e-commerce shop which three ad campaigns to kill, delivered every first Monday, $180/month, first report free." Someone who spends nothing on ads reads that and correctly says no. Someone who already has an analyst says no for a different, equally useful reason. Someone who's been meaning to fire their agency for six months stops scrolling. All three reactions are data. The vague version generated none of it.

Specificity is what turns a visitor into a signal. A page that names the customer, the mechanism, the frequency and the price is a page people can react to honestly. A page that speaks in categories — solutions, growth, empowerment — is a page people can only be polite about, and politeness is the worst possible feedback because it looks like interest.

A waiting list is not a test, it's a compliment

The instinct at this stage is almost always the same: put up a page, add an email capture, call it validation. It's worth being blunt about why that teaches you close to nothing.

An email address costs a visitor nothing. It commits them to nothing, reveals no price sensitivity, and can be given for reasons that have nothing to do with buying — curiosity, politeness, wanting to seem supportive to a friend who built the thing. A list of four hundred addresses is compatible with a business that sells to none of them and one that sells to all of them, and the page cannot tell you which world you're in.

What actually tests something is asking for a small cost. A price on the page, even a provisional one — "founding rate $49/month, increases to $79 after the first ten spots" — turns the visitor into someone deciding whether the number is worth it, and their decision is legible. A refundable deposit filters out everyone who was only being polite. A calendar link for a paid discovery call does the same thing with time instead of money. An application with two or three real qualifying questions filters harder still, and the questions people abandon on tell you almost as much as the ones they finish.

None of this requires a product, only a page willing to ask for something instead of merely collecting interest. The founder who runs a waiting list for six months and launches to four hundred subscribers who mostly don't buy has built an accurate list of people who liked the idea of her.

What to decide now, and what to leave alone

The name is worth real thought, because changing it later is expensive in a way that's easy to underestimate — every mention and screenshot carries the old one forward. Pick something that describes the category loosely enough to survive the offer changing shape, which it probably will. A name built around one specific mechanic ("The Ad-Kill Report") is a liability if the mechanic turns out wrong; a name built around the outcome or the audience survives a pivot the mechanic-specific name wouldn't.

The domain is the opposite case: worth almost no thought right now. The common failure is buying the name across six TLDs and wiring up DNS, quietly committing to a decision that was supposed to be provisional, because a domain you paid for and configured starts to feel like a decision you already made. If the exact .com is genuinely scarce and you're confident in the name, reserving it is cheap insurance. Registering it or building around it is not — a domain is a sunk-cost machine, and the point of this stage is staying free to be wrong.

Page count is the third decision, and it should also wait. One page isn't a compromise at this stage, it's correct — there's nothing yet that needs a second one. We've made the general case for right-sizing page count separately, in how many pages a small company website needs, and the pre-revenue case is the sharpest version: a site with an About page, a Services page and a Blog before the first sale is spending structure on a business that hasn't earned any yet.

Where a CV-driven generator fits, and where it stops

For exactly this stage — one page, stated plainly, live today, thrown away without regret in three months — a tool like reach fits the job better than anything built for a company that already knows what it is. It was built for a single person's page, not a business's, but the mechanics are close enough to matter: you feed it what you already have, and it produces a live draft immediately rather than an empty canvas to fill.

The number worth quoting is the real one: reach generates a complete page in about twenty seconds, and you can be live at a free subdomain in under two minutes. You still pick a colour and a hero layout from a set already filtered to your role, which takes under a minute — the real decision comes after generation, when you're looking at a finished draft and asking whether it's right, not staring at an empty canvas. The input is a CV rather than an offer statement, which is the honest limitation of using it for this job, but the output is a working, editable draft you can retitle and rewrite faster than it takes to open a page builder and stare at a blank hero section. Publishing to the subdomain is instant and free.

The limits are the reason it belongs here only briefly, and they're worth stating without softening them. It makes exactly one page — no sub-pages, no site tree. There's no CMS, because there's nothing yet that needs to grow. There's no shop and no contact form; the contact section is a mail link, which is exactly the amount of infrastructure a hypothesis needs. There's no HTML export, so you can't take the markup with you once the hypothesis survives. And you cannot connect a domain you already own — the only path to a custom domain is buying one through reach itself, a real blocker for anyone who registered their name somewhere else, and precisely why the advice above is to not register it yet. None of that is a defect here — it's the shape of a tool built to be temporary, used for something that should also be temporary.

The signal to stop testing and start building

The moment to retire the throw-away page isn't a date on a calendar, it's a change in the kind of no you're hearing. Early on, the refusals are about the offer itself — wrong price, wrong audience, wrong problem — and that's the page doing its job. The signal to move on is when the refusals stop being about the offer and start being about logistics: someone wants to pay but needs an invoice with a VAT number on it, someone wants a second package the page doesn't describe, someone asks if you also serve their industry. Those aren't objections to the hypothesis — they're evidence it held, and the page has run out of room to hold what comes next.

That's also the point where the earlier decisions you deferred stop being deferrable. The name has probably survived, because you picked one built to survive a pivot. The domain needs registering now, on your own terms, not reach's. And page count opens up for real: a business that has to answer "do you serve my industry" and "can you invoice me" has more than one job for its site to do — which is where what a service business of one should put online picks up, for exactly the founder who has just stopped needing a hypothesis and started needing a company.