My Tiny AI Tool Made $412 in Its First Month. Here Is Every Number.
A full breakdown of costs, signups, conversions and the mistakes I made launching a small paid tool as a solo builder.
Last month I launched a small tool that turns long customer support threads into a clean summary and a suggested reply. It is not a unicorn. It is a side project I built in about three weekends. But I promised to share the real numbers, so here they are, including the embarrassing ones.
The headline numbers
- Revenue: $412
- Paying customers: 23
- Price: $19 per month, with a few annual plans at a discount
- Free trial signups: 318
- Trial to paid conversion: about 7 percent
- Refunds: 2
That 7 percent number looks fine on paper. The more interesting story is where those 318 signups came from.
Where users actually came from
- A single post in a niche community for support team leads: 141 signups. This was the whole launch, honestly.
- My own posts on social media: 74 signups, mostly other builders who were curious and never intended to pay.
- A directory listing for AI tools: 61 signups, very low intent, 1 paying customer.
- Word of mouth and random search: 42 signups.
The lesson I keep relearning: one post in front of the right twenty people beats ten posts in front of a general audience. The community signups converted at almost 12 percent. My builder friends converted at under 2.
The costs
Here is where I spent money:
- Model API usage: $38
- Hosting and database: $14
- Domain and email: $3
- Payment processor fees: about $16
Total: roughly $71, so the tool is profitable in the narrowest sense. Not counting my time, which, if I am honest, was close to 60 hours.
What I got wrong
I launched without onboarding emails. Around 40 percent of trial users connected their helpdesk and then never ran a single summary. A simple day-two email showing an example output would probably have rescued some of them. That is the first thing I am adding.
I priced for builders, not buyers. Several support leads told me $19 felt suspiciously cheap for a team tool. One literally asked if there was a team plan so she could expense it. I am testing a $49 team tier next month.
I over-built settings. I spent a whole weekend on custom tone options. Usage data says 3 people have touched them.
What I would tell myself on day one
Find the room where your users already talk to each other. Show up there with something useful, not something to sell.
Next month I will share whether the onboarding emails and team pricing moved anything. If they do not, I will share that too. Building in public only works if the boring and disappointing numbers are included alongside the good ones.
Cover: Photo by Alexander Grey on Unsplash.
