The Tool That Almost Had Users

It was a Tuesday night in February, somewhere around 10:45, and I was watching a number on my screen that had not changed in nine days. Eleven sign-ups. Four sessions. One returning user, and I was pretty sure that one was John.

The tool worked. That was the part that kept me up. It wasn't broken, it wasn't embarrassing — it did the thing I built it to do. It just had eleven users, four of whom had bounced after thirty seconds, and one of whom was definitely John because he texted me the next morning to say it was "pretty clean, but also kind of sus how fast it answers." That is the full extent of my user feedback from the first two weeks.

I had told my oldest about it when I finished the build. Showed them the screen, the little interface, the way it handled the input. They watched it for about four seconds and said,

"That's cool, Dad. Can I get back to Roblox now?"

How a Boring Problem at 11 PM Became a Tool Nobody Asked For

The idea came from something I actually needed. I kept doing the same tedious lookup task for a project at work — nothing sensitive, just a repetitive thing that took three minutes every time and happened six or eight times a day. I built a small script to handle it. The script worked great. Then I thought: other people probably do this same thing. What if I wrapped it in a clean interface and put it online?

That's the thought that ruins a Tuesday. "What if I just—" is how I've started every project on my whiteboard that never got finished, and I knew that going in. But this one felt different because the core logic was already written. I wasn't speculating about whether it was possible. I had a working version sitting in a terminal window. I just needed to dress it up.

I've been down this road before — I have two tools with actual users, which sounds modest until you remember that most of the things I build never reach one. So I knew the gap between "working in a terminal" and "working for a stranger on a phone" was real, but I also knew I could close it. I spent three evenings on the UI. Nothing fancy. A text field, a button, a result. I put it on a subdomain, wrote four sentences of copy, and posted a link in two forums and one Discord server I'd been lurking in for months.

Eleven people signed up in the first four days. I thought I was onto something.

Eleven Sign-Ups, Four Sessions, and the Specific Moment the Number Stopped Moving

The first week looked like momentum if you squinted. The sign-ups trickled in over a few days — not a flood, but steady enough that I kept the analytics tab open on my second monitor while I was on work calls. I refreshed it more than I should have. I told John about it and he signed up immediately, which was generous of him but also slightly skewed my sense of the trajectory.

By day five, the sign-ups had stopped. Not slowed — stopped. The number just sat there. I went back to the forums and Discord to see if anyone had mentioned it. One person had replied to my original post with "neat," which I chose to read as positive. No one had shared it anywhere I could find. The four sessions in the analytics were John, one person who clicked around for forty seconds and left, one person who actually used the main feature and then never came back, and one session I'm still not sure was a real human.

The tool had a free tier and a "pay once for more runs" option that I'd set up mostly out of habit — I'd read enough about this approach that it felt like the right shape for something like this. Nobody hit the paywall. Not because they didn't get there, but because nobody used it enough to get there. The drop-off happened at the front door.

I did what you do when the numbers don't move: I changed the copy. Rewrote the four sentences. Added a short explainer paragraph. Tweaked the button text. Posted the link one more time in a slightly different forum. Got two more sign-ups, zero sessions. I'm not sure those two people ever loaded the page past the sign-up confirmation. This is the part I keep thinking about when I read anything about tools that almost find their second user — the gap between "someone gave you their email" and "someone actually used the thing" is enormous and I had no idea how to close it.

Three Evenings, One Domain Fee, and the Specific Tax of Knowing It Worked

Time: three evenings building the interface, probably four or five hours total. One evening on copy and posting. Two or three hours spread across two weeks of checking analytics, tweaking things, and refreshing tabs. Call it ten hours of real effort, plus another four or five hours of background anxiety that I'm not counting but also can't not count.

Money: the domain I put it on cost me nothing new because I was already paying for the hosting. The API calls behind the feature cost me somewhere around three dollars total across all the real sessions, which is either very efficient or very telling depending on how you look at it. I didn't spend anything I wouldn't have spent anyway, which sounds like a win until you remember what I could have done with those ten hours. I had DoorDash orders sitting in the queue on at least two of those evenings. I let them sit.

The real cost was the specific flavor of frustration that comes from a thing working correctly and still not finding anyone to use it. A broken tool is easy — you fix it or you don't. A working tool with no users is a different problem and I genuinely don't know enough about that problem to solve it. I've said before that the shipping part isn't the hard part, and I stand by that, but I am starting to think the part after shipping is harder than I've been giving it credit for.

I also told the brother-in-law about it at some point, which I regret. He got excited, said it had "real potential," and then spent twenty minutes explaining how I should add a referral program. I did not add a referral program. I nodded and waited for him to move on to his next scheme, which took about two more minutes.

What I Kept, What I Dropped, and Why John Is Still the Only One Who Logged Back In

The tool is still up. I didn't take it down because taking it down felt like paperwork and also because John checks in on it occasionally, which means it has one loyal user and I'm not going to zero that out over pride. I stopped actively promoting it around week three. I stopped checking the analytics daily around week four. I still check it maybe once a week, which is either hope or compulsion and I'm not sure which.

What I actually kept from the experience was the underlying script — the boring terminal version I built for myself. I still use it. It still saves me three minutes six or eight times a day. That part works great. The part where I tried to make it useful to strangers is what didn't transfer, and I think I understand why now in a way I didn't before I tried.

The thing I dropped was the assumption that "it works for me" and "it will work for someone else" are the same sentence. They're not. I built the tool for a version of me who already understood the problem, already knew what the tool was for, and had already decided they wanted it. That's not a user base. That's a mirror.

I've been thinking about this alongside some of the other almosts in the pile — the month where everything nearly clicked is a good comparison point, because that one also had a moment where the numbers looked like they were going somewhere and then just didn't. The pattern is familiar enough that I should probably write it on the whiteboard. I won't, but I should.

John texted me last week to say the tool gave him a weird answer on one query. I told him I'd look into it. I have not yet looked into it.

Thirteen sign-ups. Maybe five real sessions. One guy who came back, and he's my friend, and he filed a bug report via iMessage. The tool works. The number just doesn't move. I don't know what to do with that yet, and I'm not sure I have the hours to figure it out before the next thing pulls my attention somewhere else.

Note: Everyday Hustle Jeff is the written side of a digital series. These accounts are composite and dramatized. Any figures are what they were on the day, for one person, and are not a projection, a track record, or advice.