Launched, and nobody came

You shipped. You posted it somewhere. There was a small spike, some polite congratulations, and then a flat line. It has been three weeks.

The instinct at this point is to conclude that you need marketing, and marketing is a big vague thing, so you go back to the codebase where problems have solutions. That is the actual risk of this moment, more than any wasted spend.

First, split the problem in two

“No users” is two conditions that feel identical and want opposite treatments.

• Nobody is arriving. Your page is fine and no human has seen it. This is a distribution problem.
• People arrive and leave. Traffic exists and converts nothing. This is a clarity problem.

You cannot tell which from feeling, so go and look. Any analytics will do. If you have none installed, that is your next twenty minutes, and note that it means every decision you have made since launch was made blind.

If nobody is arriving

This is the more common case and the better news, because it is fixable this week by hand.

The launch was not distribution

Posting once on a directory, a subreddit, or a launch site is an event, not a channel. Events produce a spike and a flat line, which is exactly the shape you are looking at. Nothing has gone wrong; you simply have not done distribution yet.

Do the unscalable thing on purpose

Find twenty people who have the problem, in public, in the last month. Not twenty people in your target market. Twenty people who wrote something that shows they have the problem.

Search the language of the complaint rather than the language of the category: “anyone found a better way to”, “still doing this manually”, “how do you all handle”. Then be useful to each of them individually, with no link in the first message.

Twenty conversations tells you more in a week than twenty thousand impressions.

This is slow, it does not scale, and it is the only reliable way to start. Getting the first users goes into the mechanics.

Where to turn up, without being a nuisance

The worry that stops most founders here is a real one: nobody wants to be the person who joins a community to advertise. The distinction that keeps you on the right side of it is simple and it is not about disclosure or tone.

If your comment would still be worth reading with the product name removed, post it. If removing the name makes it pointless, do not.

• Answer questions rather than announcing things. A thread where somebody asks how to handle your problem is an invitation. A thread about something else is not.
• Sort by recent. A question from last week has a person attached who still has the problem. A popular thread from 2021 has an audience but no live need.
• One room at a time, for a month. Posting the same link across twenty communities in a day is the one move that costs you all twenty at once.
• Read the rules, genuinely. Most communities have a self-promotion policy, and many have a designated thread or day for it. Using it is both allowed and much more effective than being removed.

The boring option, which works

Directory and listing surfaces are unglamorous, deterministic and finite: you work through a list, each entry takes ten minutes, and each one either accepts you or does not. No judgement required and no risk of annoying anybody.

The traffic per listing is small and the total is not nothing, the pages persist, and being described in a directory forces you to write the one sentence about your product that you have been avoiding. It is also the right thing to do on a day when you have twenty minutes and no energy, which is most days.

If people arrive and leave

Harder to accept, faster to fix. Something on the page is failing in the first few seconds.

The three second test

Send your URL to three people who do not know what you built and ask one question: what does this do, and who is it for? Do not explain first, do not watch over their shoulder, do not defend the answer.

Most technical founders fail this badly, and it is a specific failure rather than bad writing. You describe the mechanism because the mechanism is what you spent six months on. The visitor needs the outcome and the recognition of their own situation. “Realtime sync engine with conflict resolution” against “your team stops overwriting each other’s work”.

Check the path, not just the words

• Can somebody try it without a signup, or is the first thing you ask for a credit card?
• Is there one obvious next action above the fold, or four competing ones?
• Does the empty state on first login tell somebody what to do, or does it show them a dashboard with nothing in it?
• Does it work on a phone? Actually check, on a phone.

What not to do this month

• Another feature. The most expensive mistake available to you, and the most comfortable.
• A redesign. Different from a rewrite of what the page says. The words are the problem, not the layout.
• A content programme. Six to twelve months to pay back, and you need to know something now.
• Paid ads. Buying traffic to a page that does not convert is paying to confirm the diagnosis you could have got free.
• Relaunching. You get one launch of attention per audience, roughly, and doing it again with the same product is spending goodwill for another spike and another flat line.

The one thing for this week

Pick whichever half of the split you landed in, and do the single action for it. If nobody is arriving: twenty individual, specific, link-free messages to people who have publicly described the problem. If people arrive and leave: rewrite the first two lines of your homepage in the words your three testers used when they got it wrong.

One of those, this week, measured. Not both, and not a plan containing both. If you would rather have that judgement made against your actual site, the Product Check is free and needs no account.