For three years I thought building software worked like this: I get an idea, I build it, I ship it, and let's see what it catches.
Eight projects, none of them finished, zero users.
And I spent all of it sitting at my desk, sitting on the couch, asking an LLM, reading forums, looking for validation inside documents, and I honestly thought that was research.
I had my ideal customer perfectly defined, the problem is that this customer existed under my own description, with the characteristics I gave him myself, and the ideal customer isn't in a document or a text file or a social media post, he's out in the field, he's on the street.
Four months ago I went out to find him.
I broke
The first time I went knocking on doors I broke, I didn't even want to do it, I felt nauseous before getting out of the car.
The first three or four conversations were a disaster, I couldn't form the words, the ideas wouldn't come either, I'd forget what I was asking them for in the first place, and this is with a script in my hand, but I'd lose that same script by the second answer.
It took me about eleven weeks to get that off my back, and then at some point it just stopped happening, I'd walk in and already know what to ask, I already had the information from the last one, I already knew what the next question was, it turned into routine.
I knocked on around 50 veterinary clinics in total, not to sell them anything, but to understand what hurt and how they were solving it today.
And what I found I didn't like: almost all of them had already solved it, not perfectly, but solved, they already had their own solution, their own patch, their own pain handled on their own, and there I was showing up wanting to sell them something of mine they never asked for.
The first time I said a price
There was one clinic I visited four times, four, I already knew the owner, we'd already talked a bunch of times.
With what I'd learned from the other 50 I put together a proposal and went to present it, and when the moment came to say the number, I couldn't, I was shaking, my voice cut out, and I felt a panic I've never felt in any deploy or any technical interview.
She asked for time, and that was that.
I lost a client without having written a single line of code, all of it had been conversations.
It sounds bad, but that's exactly the good part: losing there is cheap, because before, losing cost me three months building something nobody was going to open.
The second time it came out whole
Then I got to a bigger clinic, she looked at the proposal, said she'd like to try it, and that's where I asked a second time, except since it was a bigger company saying it in person wasn't enough, I had to send it in writing, through formal channels, with all the information.
That time the question came out whole, a deposit, with a hard deadline, nothing watered down, and in years of projects I had never done that.
The answer was: more time, and then more time again, and then more time again.
Three weeks in I already knew that person wasn't interested, that she was just stalling, so I set a limit and it died there.
Eleven sprints in that vertical, zero clients, zero pesos, zero letters of intent.
I kept asking for a while longer, changed audiences, changed the angle, but I kept hitting the same dead end: volume was there, there are plenty of vet clinics, but when it came to asking for a deposit, even one with a 100% refund, nobody would commit.
Failing for four months is expensive
That's where I stopped to think, because failing for three or four months is expensive, as a developer and as a person those are months of my life, and you don't get them back.
So instead of just switching ideas, I switched the question: how fast can I find out this isn't going to work?
I built a spreadsheet with criteria to pick the next niche, three finalists came out, I took the first one, and went back to asking in the field, because I wasn't going to build anything without a confirmed need, if I don't have the need I don't even know what they're looking for.
And going out again taught me things no forum was going to tell me.
There are mobile businesses and non-mobile businesses, and I didn't know that, I went to the addresses and they simply didn't exist, I walked the whole block and it was all residential, no sign, not a single clue, and they still had published hours, a phone number, a website and around twenty reviews each, which means the entire industry was working out of their houses.
And then there are the ghost businesses, the listing said more than 50, but the ones actually working were fewer than ten.
Those ten already had systems behind them and I couldn't get much more out of them, and of the few that did answer me almost all turned out to be side hustles, their sales cycle was extremely long, their volume was low, and for what I was going to charge for a SaaS it just wasn't worth it to them.
That's where the question I now always ask came from, the one I wish somebody had asked me three years ago:
How many customers does that person have to serve just to pay for my subscription?
Not what my product is worth, but how much the other person has to bill for my price to make sense inside their business, and whether I'm saving them anything, how much I'm saving them, and whether it's money or time.
If the answer is "way more customers than they actually get", there's no product, no matter how well built it is.
I didn't kill that vertical with a conversation, I killed it by counting, the market was too thin and you could see that on a map, not in an interview.
Two weeks, and the previous one had taken me three months.
Where I am
These four months, in numbers: 50 businesses visited in person, more than 40 calls, more than 80 people I talked to about what hurts, 10 times I asked for a price, and 89 nos.
And with that I built a machine that actually works: pick a niche with criteria, go out, ask, measure, and kill it fast when it doesn't work.
That machine has been producing zero revenue for four months.
Both of those are true and neither cancels the other out, and I'm not going to write here that the real product was the learning, because no, I still don't have a client.
The only thing I can honestly measure is this: the last time I asked for money it was over the phone, I said the full number, I asked for the commitment, I asked the closing question, and then I shut up, I didn't fill the silence, I didn't add "if that's too much we can figure something out".
For someone who a few months ago couldn't even finish the sentence, that's what changed.
I still don't have a paying client, but for the first time in three years the problem isn't me anymore, and that's something you can actually work on with a method.
