If you shipped something that works and nobody uses it, the bottleneck is almost never the code. Roughly two thirds of solo-built products that fail cite distribution as the cause. Distribution is a system with three stages — knowing who has the problem, reaching enough of them, and saying something they answer — and each stage fails in a way you can diagnose. Launch days are not distribution; they are one impulse in a system that needs a steady signal.
It works. You tested it on yourself for weeks. You fixed the edge cases, you wrote the docs, you paid for the domain. You posted it. Some people said nice things. Then the graph went flat and stayed flat, and you have been carrying a specific kind of quiet embarrassment ever since.
Here is the first useful thing to know: this is the most common outcome, and it is not a verdict on the product. A widely cited figure from the indie community puts it at roughly two thirds of solo-built products that fail to get traction naming distribution — not the code, not the pricing, not the market — as the reason. You are in the majority, and the majority is failing at the same specific thing.
Shipping stopped being the hard part
For most of software history, the constraint was building. If you could build the thing, you had something rare, and being rare did a lot of the selling for you. That is over. Tooling, frameworks and AI have collapsed the cost of building to the point where a competent person can ship a real product in a weekend, which means shipping no longer distinguishes you from anyone.
What did not get easier is the part that comes after. Getting a stranger to notice, believe and pay is the same difficult human problem it has always been, and nothing about being good at engineering prepares you for it. The advice you were given — build something people want — quietly assumed that people would find out it existed.
Where it actually breaks
Distribution is not one thing that either works or does not. It is three stages in sequence, and a flat graph tells you one of them is broken without telling you which. Founders burn months here because they change tactics at random instead of locating the fault.
Stage one: you do not know who has the problem
Not the segment. The people. "Startups" is not an audience. "Series A fintech companies in the US who just posted a compliance engineering role" is an audience, because you can enumerate it, and enumerating it is what makes everything downstream possible.
The tell for this failure: you cannot produce a list of two hundred specific companies that have the problem you solved. If you cannot write down the list, you have been marketing at an idea of a customer rather than at customers. A concrete method for getting from product to list is in the ICP guide.
Stage two: you are reaching too few of them
This is the stage that quietly kills almost everything, because the numbers are counterintuitive. A good cold email reply rate is somewhere between two and five percent. A good reply-to-meeting rate is maybe a third. So thirty meaningful conversations — enough to learn whether the thing sells — needs something on the order of two to three thousand well-targeted messages.
Most builders send forty, hear nothing, and conclude outbound does not work. Forty is not a test. Forty is noise. The maths says nothing at all should have happened, and nothing is exactly what did.
| Stage | Rate | You need |
|---|---|---|
| People contacted | — | 2,000 |
| Replies | 2–5% | 40–100 |
| Conversations | ~1 in 3 replies | 15–35 |
| Sent per day, safely | ~25 per address | a schedule, not a weekend |
Stage three: the message gives them no reason
The third failure is the one that feels most personal. You reached the right people, enough of them, and they ignored you — because what you sent was about you. Most first cold emails are a description of a product written by the person who is proud of it, sent to someone who has never heard of it and owes it nothing.
The messages that get answered reference something specific and true about the recipient, then connect it to a problem they already know they have. That is not a tone or a template. It is research, done per person, at a volume no human wants to do by hand.
Why launches feel like rejection
Product Hunt, Hacker News, a Reddit post, a Show HN. These are worth doing, and they are not distribution. They are impulses: a day of traffic and a week of silence, and the silence after a spike reads emotionally like rejection when it is just the shape of a spike.
A distribution system produces a steady signal rather than an impulse. It runs on days when you are not thinking about it, it does not depend on being featured, and it does not require an audience you have not spent two years building. Outbound is the only channel with those properties available on day one, which is the entire and unglamorous reason it is the right first move.
The part nobody mentions: it is mostly maintenance
Outbound done properly is not a burst of effort. It is infrastructure that needs tending. Sending domains that are not your real one, warmed over about three weeks before they carry anything. Daily volume held under a limit per address. Lists refreshed. Messages rewritten when reply rates move. Suppression for anyone who asked you to stop.
This is where most builders stop, and it is a rational place to stop, because the work is repetitive and specific and you did not start a company to do it. It is also the work that is most completely automatable — not the writing-a-sentence part that AI demos always show, but the whole pipeline underneath it: choosing who, verifying, sequencing, throttling, watching reply rates, adjusting.
That is what Operater's Sales agent is: the infrastructure and the judgement, not a text box you have to brief. It reads what you built, works out who has the problem, buys and warms its own domains, writes per person, sends on a schedule that protects your reputation, and changes the approach when the numbers move. You do not describe the job, because the whole premise is that you have never done the job.
What to do this week
- Write down two hundred specific companies that have the problem you solved. If you cannot, your fault is at stage one and nothing downstream will help.
- Pick the thirty most obviously affected and contact them by hand. Badly is fine. You are buying information, not conversions.
- Read every reply, including the rude ones, and note the objection you did not anticipate. That sentence is worth more than the week.
- Decide whether the numbers justify doing this at two thousand rather than thirty. If they do, stop doing it by hand, because two thousand done by hand is a job you will quit.
The product is probably fine. The distance between shipped and used is a system you have not built yet, and unlike the thing you already built, this one has been built before, many times, by people who wrote down how.
Key takeaways
- Shipping is no longer the hard part, and has not been for several years. Distribution is the constraint, and it is a different discipline you were never taught.
- The failure has three possible locations — wrong audience, too few of them, or the wrong message — and they need different fixes. Guessing which one is broken is why founders churn through tactics.
- A launch is a spike, not a system. Product Hunt and Hacker News produce a day of traffic and a week of silence, which reads as rejection when it is just the shape of a spike.
- Outbound is the only channel where you control the volume, which is exactly what makes it the right first channel when you have no audience.
- The work is repetitive, specific and unglamorous, which is why it is the part most builders avoid and the part that most rewards being automated properly.