I killed my first product
It was called HowsThisGoing (@howsthisgng) // howsthisgoing.com, and it was my first attempt at building something entirely on my own, from scratch.
In ~2.5 years, it made me ~$400.
I probably should've shut it down way earlier. But I kept coming back to it thinking "oh, I'll make that one change and then it'll make money."
It never did. I don't regret any of it, though. And I think I finally understand why it never worked.
It was before everything
I started building it about two months after leaving my first job. I'd been a Product Manager at a company that taught me a lot about how to build and ship things, and now I just wanted to code & build something of my own.
Anything. Hopefully something that made money.
The problem was, I couldn't really code. The only code I'd touched was HTML/CSS, SQL (lol, ik it's not really code) and some basic C and Python from Harvard's CS50.
This was the early ChatGPT era. Cursor was just getting started, and Claude Code, Codex and most of the tools I use every day now didn't exist yet. I didn't even know how to use my Mac's terminal. I remember asking ChatGPT how to cd into a folder. How to mkdir one.
Before writing my first real line of code, I went through the Django tutorial on their official site. It taught me apps, models and basic CRUD, and I remember thinking: man, this CRUD thing can run a lot of cool shit.
Then I had to buy and set up an AWS server. I'd never SSH-ed into anything in my life. I was mostly tinkering, failing, tinkering again, and somehow getting things to work. I'm fairly certain it was the most fun I'd had in years.
I also had no idea what I was doing, which led to some hilarious mistakes.
- I didn't know webhooks existed, so when someone answered their standup in Slack, I had a poller running in the background, constantly checking whether any new messages had come in lmao.
- I had to take multiple classes from a friend, the CTO at my previous company, before I understood "why do I need classes in python, why can't this just be a function."
- Memory leaks everywhere. Celery, the thing that sent out every standup, kept crashing on its own. When I added Sentry to tell me when things crashed, Sentry made Celery crash too. It finally got fixed while I was setting up PostHog, of all things.
- I ran a real company's daily standups on AWS's free tier. Some days, the standups just didn't go out, and I only found out by chance.
For a sense of how much has changed since then: at one point I was paying around $100 a month for AWS, for a product that made $400 in total. When I finally moved it to Hetzner, it took one session with Claude, and the bill dropped to under $20 a month.
My best customer never paid
My first customer was my ex-company. They used a standup tool called Standuply, and within a couple of months I'd built something that did pretty much everything Standuply did.
I offered them a pilot, free. Once it was stable, my ex-boss wanted to switch their billing over to me. I said I didn't want to charge them.
The reply: "nice of you to offer. but it's cool we'll pay. i want to be your first paid customer lol"
At 2am, I replied: "Sweet, but yeah I've already told people that I'm not going to charge you guys. It's my first product. You could pay for the next one."
They ended up running their standups on it every day for over two years. They never paid, because I never let them.
My best customer was a team I'd refused money from.
I don't regret it. I learnt a ton from those guys, and it was my way of giving back. But it took me a long time to see what I'd given up.
They still shared feedback. But if they'd been paying, they'd probably have had a reason to push, and I'd have had a reason to listen. Instead, I'd turned their feedback into a favour, and that was on me.
Shooting in the dark
So I kept building, mostly on my own guesses.
Teams could answer standups with files, images and docs, not just text. Then I added embeddings across all of it, so you could ask the bot what got done, what was next, who was blocked. It could set up and run standups on its own, right from Slack. It had RAG and memory before anyone I knew was really talking about either.
One of the few features that didn't come from my own head came from a random piece of feedback. An engineer told me, "as an engineer, I hate posting what I worked on."
So I built a GitHub integration that did it for them. It turned what the team actually shipped into daily, weekly and monthly snapshots, so nobody had to write it all up. I built the architecture so Linear, HubSpot and others could plug in the same way... but I lost motivation before I got to them.
#6 of the day
For a while, people noticed.
- Launched on Product Hunt and finished #6 of the day.
- Posted a Show HN and got a few users for the first version, including someone from an Indian unicorn.
- An AI newsletter with over 100k readers featured it.
- There's An AI For That (TAAFT) ranked it #2 in team management, with over 9 million search impressions.
It felt like it was working.
It wasn't. The launches brought people in. They didn't bring in customers.
Free to leave
Then I made the same mistake again, this time with everyone.
HowsThisGoing had a free plan. Small teams, under ten people or so, could run four or five standups without paying anything. Paid plans started at $80 a month.
So people came in, tried it, decided it wasn't for them, and left. Quietly. No payment, no cancellation, no reason.
And I never spoke to any of them. Which, looking back, is kind of fucked up. People had found my product, signed up and given it a shot, and I never once asked them why they left. I just wasn't motivated to.
If they'd paid and then asked for a refund, at least I'd have had something. An email, a reason, some kind of loop forming. A refund request is a conversation. A free user leaving is silence.
The free plan should never have been there.
The traffic died
When the launch traffic dried up, I tried to automate my way into customers. I set up Apollo and Instantly and ran cold email campaigns. Nothing. I tried automated LinkedIn outreach. Nothing.
I was cold emailing strangers, while the people who'd already tried my product never heard from me.
So I went back to the thing that felt productive: building. Instead of talking to team managers and CEOs to find out what they actually needed, I was setting up daily database snapshots so everyone's data would be safe once I had thousands of users. I was building for users I didn't have.
Instead of doing outreach by hand, I was hacking around the Slack connection. There was a lot more like that. I kept doing the things I thought I was supposed to do, instead of the one thing I actually needed to do - talk to users.
Never my problem
It took me way too long to see the real problem: I wasn't close to it.
I wasn't the person who had to make standups work. I'd never managed a team, especially a large one. I'd seen the problem from the side at my old job, and that was about it. So I didn't know who to talk to, or what to ask them. Every cold email I sent was a guess at someone else's pain.
The one time I was close to a real user's pain was an engineer who hated posting what they'd worked on. That turned into the GitHub integration, and it was the best thing in the product.
But slowly, I lost motivation, because it had never really been my problem.
Too long, and too early
The standard indie hacker advice is: if it doesn't print money, let it go.
I bought into it. I told myself "this will never make money, let's build something else," and moved on.
I'm not sure that was right either. The companies I admire weren't built by people who moved on when the first version didn't print money. They were built by people who spent years making one thing work.
Today there's a whole wave of AI teammate apps, like Viktor, doing a lot of what I was trying to build back then.
So I held on to the product for too long, and gave up on the problem too early. It sounds like a contradiction, but it's really the same mistake. I could only have spent years on it if it was a problem I was close to, and it never was.
Tuition
Building all of this is how I figured out what I actually want to spend years on: travel. That's what I'm building now with @pointscasa.
I'd love to say I've learnt my lesson completely. I haven't. I still catch myself treating it like an engineering problem, tinkering and hacking around with things. But this time, I know exactly who I'm building for, and I talk to them often.
HowsThisGoing is off now, but nothing's been deleted. If you ever used it and want your data, DM me. And thank you to everyone who used it, especially my ex-company, for running their standups on it every day for over two years.
$400 is a terrible return on 2.5 years.
But as tuition, it was the cheapest I'll ever pay.
I don't post often, but when I do it's worth your time. No spam.