The first $100 MRR: eight weeks, four features, one refund

Eight weeks after launch the tool is at $103 MRR. Not a rocketship. Not a failure. Just enough signal to write down what I got right and what I got wrong.

Eight weeks after launch the tool is at $103 MRR — twelve paying customers, one refund, roughly $6.20 in Stripe fees and $19 in server costs. Not a rocketship. Not a failure. Just enough signal to write things down before I forget.

The rest of this post is a plain accounting: what actually moved the number, what did not, and what I would do differently if I were starting again next week — which, given the way things go around here, I might well be.

What the product actually is

A very small tool that takes a CSV of receipts and produces a monthly summary in the format my accountant asks for. It exists because I got tired of copy-pasting the same three columns four Fridays a month. I built it in a weekend, shipped it on a Sunday, and did not tell anyone for three weeks because I was too embarrassed by the landing page.

Pricing: $9/month. No free tier. Fourteen-day money-back guarantee. That is the whole business.

Weeks 1-2: nothing happened

I put the site up. I told nobody. Google eventually indexed the landing page. Three people found it. Zero converted. This is the correct outcome for that level of effort, and I include it here only because founders tend to talk about "launch" as if the moment of pushing to production had any weight on its own. It has none.

Week 3: the first paying customer

I posted a small write-up on a niche forum for indie accountants that has, generously, about 400 active members. I described the problem, showed a screenshot, and linked to the site. I did not use the word launch. I did not ask for feedback. I just described the thing.

Fourteen visitors. One signup. One conversion — at midnight local time. I refreshed Stripe about eleven times before I believed it.

Nine dollars in your Stripe dashboard, from a real person you have never met, is not a small feeling. It is worth quite a lot of subsequent grinding.

Weeks 4-5: figuring out what people wanted

The next four customers came from that same thread and a follow-up post I made a week later. In their onboarding emails I asked two questions:

• What did you paste into Google before you found this tool?
• What is the smallest thing about this tool that would make you cancel?

Every single answer was useful. Nobody said "I want more features." Everybody said some variant of "please make the CSV format less strict" and "please give me the export as a PDF too." One person said "please stop sending me a receipt email, it's noise." I noted this and did nothing for two more days, then fixed all three in a weekend.

The result of fixing them: three of the four filed at least one more successful summary that week without emailing me for help. This is, I think, the definition of product-market fit at this stage — not that people love you, but that they stop needing you to be in the loop.

Week 6: the refund

A very kind person from a country whose banking format I had not tested at all uploaded a CSV, got garbage back, wrote to me apologising for the inconvenience, and asked for a refund. I gave it to her within nine minutes and thanked her. She thanked me back and offered to send me sample files so I could fix it. I told her I would; I did, that weekend. She is not (yet) a customer again, but she recommended the tool in a comment on the forum thread, which sent me two more signups.

The lesson: refund quickly, thank the person, and follow up on the root cause. There is nothing else to say about this and I do not know why some founders make it so complicated.

Week 7-8: the boring, effective stuff

The last two weeks of MRR growth came from three things, none of them clever:

1. Rewriting the landing page. The old one had five feature bullets and no example. The new one has one screenshot, one paragraph, one price, and a big "try it on your own CSV" box. Conversion roughly doubled.
2. Adding a "sample CSV" download so the first-time visitor could see, in three seconds, what shape the tool expected. Support emails dropped by a third.
3. Setting up a tiny email sequence for people who signed up but never uploaded a file. Two of the twelve current customers came from this email — meaning about 17% of MRR is directly attributable to a 20-line python script.

What did nothing

• Twitter. Zero attributable signups. I tried, several times, and got exactly one polite quote-tweet from a friend.
• A "Product Hunt" listing. About 40 pageviews, zero conversions, one email from an SEO agency asking to buy backlinks.
• Paid search. I burned $22 on a Google Ads test with no conversions before I turned it off. My CPC on the phrase I actually wanted was $6.40, which does not work for a $9/month product with (currently) a five-month median lifetime.

What I would do differently

Ship the tool exactly as I did — small, ugly, priced from day one. Do not build a free tier. Do not build "an API" before there is a customer asking for one. Do write the onboarding email sequence in week one, not week five. Do put a real screenshot on the landing page before "launching."

And — this is the one that hurts — do not use the phrase MVP. It puts you into a mindset where "minimum" is the ceiling. Ship the small tool as the tool. It will get better, or it will not, but it is not a preview of anything.