Most advice about getting your first hundred users assumes you have something most post-launch founders do not have. An audience. A newsletter list. A Twitter following that has been warming up for two years. If you have those, the playbook is easy: tell the people who already trust you, and some of them convert.
This post is for the other founder. The one who built something good, shipped it, and then realised the hard part was only just beginning. You have no audience, no budget for an agency, and a creeping suspicion that growth is a thing other people are mysteriously better at.
It is not mysterious. Early growth is a small number of mechanical things done with more consistency than feels reasonable. Here is how the first 100 users actually arrive, and why almost nobody talks about it honestly.
- The launch myth treats day one as the finish line. Your first 100 users come from steady discoverability, not a single spike.
- Users find post-launch SaaS through AI search, traditional search, communities and directories, so be present and accurate in all four.
- Consistency compounds. A discoverability flywheel that turns every week beats a one-off launch that fades in days.
- Measure qualified arrivals and what they do, not vanity traffic from launch day.
Why the launch myth fails you
The founder origin stories you read are survivorship bias wearing a narrative. The Product Hunt launch that hit number one. The Show HN that pulled thousands of visitors in a day. The tweet that went off. These stories get told because they are rare enough to be interesting, which is precisely why they are a bad model to plan around.
A launch is a spike. Spikes have a shape: they go up fast, and they come down faster. You get a flood of curious people, a handful sign up, most never return, and a week later your traffic graph looks like nothing happened. If you organised your entire growth plan around the spike, you are now staring at a flat line and wondering what you did wrong. You did nothing wrong. You just bet on the wrong shape.
The founders who get to a hundred users and then a thousand are not the ones with the best launch. They are the ones who built something that keeps getting found after the launch is over. The launch, if it works at all, buys you a little oxygen. What you do with the months on either side of it is the actual game.
Where your first users actually look
Before you can be found, you have to know where people look. A pre-traction B2B founder usually has an instinct here that is a decade out of date, shaped by how they themselves discovered software back when Google was the only front door. The front doors have multiplied.
When someone has a problem your product solves, they do some combination of the following, often in the same afternoon, without thinking of it as research.
They ask an AI assistant
A growing share of "what should I use for X" questions now go to ChatGPT, Claude, Gemini, or Perplexity before they ever reach a search box. The person does not want ten blue links. They want a shortlist with reasons. The assistant gives them three or four named tools and a sentence on each. If your product is not in that answer, you do not exist for that person. They will not scroll past the answer to find you, because there is nothing to scroll. The answer is the whole interaction.
This is the quietest shift in early-stage discovery and the one founders underweight the most. You can rank perfectly in Google and still be absent from the AI answer that sits above the search results, because the engines decide what to name by reading the wider web, not just your homepage.
They still search, but differently
Traditional search has not died. It has narrowed. People still type problems into Google, but increasingly the top of the page answers them directly through an AI Overview, and the click only happens when they want to go deeper. So you are competing for two things at once now: the ranking, and whether the synthesised answer mentions you by name. The second is downstream of being cited across enough credible pages that the engine treats you as a real option.
They read communities and ask peers
Your ICP reads Hacker News, Indie Hackers, and a handful of subreddits for their niche. When they have a question shaped like a purchase, they search those communities, or they post and wait. A two-year-old Reddit thread where someone asked "best tool for X" and three people answered is a piece of discovery infrastructure that compounds quietly for years. Communities are where pre-traction founders can earn the most attention for the least money, and where most of them behave so badly that they get banned instead.
They check directories and review sites
Product Hunt, G2, Capterra, the niche "awesome" lists, the category roundups some blogger wrote two years ago and still updates. These pages rank, they get cited by AI engines, and they carry the social proof a brand-new product cannot generate on its own. A founder with no reviews and no listings is invisible on a whole surface that customers actively trust, precisely because it is not the founder talking.
The pattern across all four: your first users are not waiting on your homepage. They are out in the world, asking questions on surfaces you do not control, and deciding based on whatever shows up. Getting found means showing up on those surfaces, repeatedly, in the user's own context.
The discoverability flywheel
Here is the part that makes the work worth doing. These surfaces are not independent. They feed each other, and that feedback loop is the closest thing early growth has to a free lunch.
Walk one example through. You write a genuinely useful answer to a recurring question in your category and publish it on your site. A few people find it through search. One of them links to it from a community thread because it actually helped. That thread ranks. A directory or roundup picks up your product because it now has a credible page behind it. An AI assistant, reading across your site, the thread, and the directory, starts naming you in answers because three independent sources corroborate that you exist and solve this specific problem.
Notice what happened. One piece of honest work seeded mentions on four surfaces, and the mentions reinforced one another. That is the flywheel. Each credible mention makes the next one easier, because both search engines and AI assistants are, underneath, trying to answer the same question: is this a real thing that real people vouch for. Every surface you appear on is another vote, and the votes are read together.
The flywheel is slow to start and hard to stop once it turns. That asymmetry is the whole reason to prefer it over launches. A spike gives you a number today and nothing tomorrow. A turning flywheel gives you a little today and slightly more every week, indefinitely, because the mentions you earned last quarter are still being read.
For the deeper mechanics of how AI engines decide who to name, see the new playbook for the AI search era. →Why consistency beats launches
If the flywheel is the mechanism, consistency is the input that drives it. This is the least glamorous sentence in growth and the truest. The founders who win early are not smarter or louder. They show up on the same surfaces, week after week, long after it stopped feeling like it was working.
There is a real reason for this beyond grit. Discovery surfaces reward presence over time, not bursts. An AI assistant names the products it sees mentioned consistently across many sources, because consistency reads as legitimacy. A community trusts the founder who has been genuinely helpful in twenty threads, not the one who showed up once to drop a link. Search rewards the site that publishes useful pages on a steady cadence, because that pattern looks like a real ongoing concern rather than an abandoned project.
Consistency also fixes the thing that actually kills early-stage growth, which is not failure but quitting. Most founders try a channel for two weeks, see no signups, and conclude the channel is dead. The channel was not dead. Two weeks is roughly the warm-up. The work was compounding, invisibly, right up until the founder stopped and reset the counter to zero.
A realistic cadence for one person
You cannot be everywhere. You can be somewhere, reliably. A workable weekly rhythm for a solo founder looks roughly like this: publish one honest, specific piece of writing that answers a real question in your category. Spend genuine time being useful in one or two communities where your ICP already gathers, with no pitch attached. Make sure your product is listed and current on the two or three directories that matter for your niche. Check, every so often, whether AI assistants name you when asked what to use for your problem, and whether they describe you correctly.
That is it. Four threads of work, repeated. None of them produces a spike. All of them feed the flywheel. The compounding happens in the boring middle, which is exactly why most people never see it.
Measure the right thing
The trap with slow, compounding work is that the obvious metric, signups today, stays near zero long enough to make you quit. You need leading indicators that turn before the lagging ones do, so you can tell the difference between work that is compounding and work that is genuinely going nowhere.
Watch whether the surfaces themselves are filling in. Are AI assistants starting to name you when asked about your category, and getting your description right. Is a community thread you contributed to now ranking and sending the occasional visitor. Are your published pages accumulating impressions in Search Console even before the clicks arrive, which is the unmistakable signal that you are being seen upstream of the click. Are you appearing on the directories and roundups your customers check.
These indicators turn weeks before signups do. If they are trending up, the flywheel is turning and you should keep going. If they are flat after a real, sustained effort, the message is not landing and you can change it deliberately rather than abandon the whole approach in a panic. Either way you are making decisions from evidence instead of from the anxiety of a flat signup count.
If you want to know specifically why a competitor keeps getting named in AI answers and you do not, this breaks down the mechanism. →The honest version of the first 100 users playbook
Your first hundred users do not arrive in a flood. They arrive a few at a time, found by people who were out in the world asking a question and saw your name show up on a surface they already trusted. An AI assistant that named you. A community thread that vouched for you. A directory that listed you. A search result that answered honestly and led them in.
The work is to be present on those surfaces, in the founder's own voice, consistently enough that the flywheel starts turning. It is unglamorous and it is slow at the start, and it is also the only thing that keeps producing users after the launch spike has faded to nothing. Builders are usually good at the product and lost at this part, not because it is hard, but because it asks for steadiness across half a dozen surfaces at once, and there are only so many hours in a founder's week.
That distribution problem is the one AfterLaunch is built to carry. It watches where users decide, from AI assistants to search to communities to directories, drafts the work in your voice for you to review and approve, and proves what changed through your analytics and rank tracking so you can see the flywheel turning instead of taking it on faith. The free Growth Snapshot is a reasonable place to start: it scores how discoverable you are across seven dimensions today, which tells you which surface to work on first instead of trying to do all of them at once.
Start anywhere. Pick one surface, show up on it every week, and let the mentions begin to feed each other. A hundred users is not a launch. It is a flywheel you turned for long enough that it started turning itself.
I launched and got a traffic spike, but it died within days. What went wrong?
Nothing unusual. A launch concentrates attention into a short window, then the audience moves on and the spike fades. The fix is to build sources of ongoing discovery so people keep finding you after the launch crowd has gone. That means being present and accurate where users actually look the rest of the year.
Where do my first 100 users actually come from if not the launch?
Usually a mix: AI assistants answering questions about your category, traditional search for the problem you solve, communities where people discuss that problem, and directories users browse when comparing tools. No single channel carries the whole load. The point of the playbook is to be findable in all of them rather than betting everything on one launch day.
How long before the discoverability flywheel actually starts working?
It is gradual rather than instant, which is exactly why launches feel more satisfying in the moment. Search and AI visibility build as content accumulates and gets cited, and community presence builds as you keep showing up. We would rather set the honest expectation of weeks of steady work than promise an overnight result we cannot back up.
Should I just keep doing launches on different platforms instead?
Repeated launches can help, but each one is still a spike that fades. The deeper return comes from the assets a launch can seed: pages that rank, answers AI tools cite, and a track record in the communities your users trust. Treat launches as one input to the flywheel, not the engine itself.
What should I measure to know the first 100 users playbook is working?
Track qualified arrivals and what they do once they land, not raw launch day traffic. Signups, activation and which sources bring people who stick tell you whether discovery is reaching the right users. Vanity numbers from a single spike will not tell you if the flywheel is turning.