zero paying customers. here's everything i got wrong.

on july 20th, 2026, my last paying customer churned.

the homepage of this site shows my MRR live, pulled from a revenue-verified listing. today it reads $0. i shipped that counter the same day the number hit zero. that's either the dumbest thing i've done as a founder or the most honest, and i think the answer is both.

this is the post-mortem. real numbers, no soft lighting.

TL;DR

• $152 in revenue from 8 paying customers. vibedraft peaked at 9 paying subscribers and $171 MRR in late may, then bled to $0 over the next two months. the last one left july 20, 2026.
• the product wasn't the problem. the funnel was. i shipped features for months while the site had no email capture and no path from reader to customer.
• 90.8% of my 1,541 analyzed posts earned zero bookmarks. i was posting into the void and calling it distribution.
• the fix isn't motivation, it's instrumentation: public numbers, kill criteria, and a monthly checklist that catches these mistakes while they're cheap.

the numbers, all of them

the revenue numbers come straight from stripe (my own two test accounts excluded), and the follower count on my homepage is pulled live from the x api. i'm done grading my own homework.

mistake 1: i already knew this lesson, then unlearned it

my first startup was a college social network called studentgrounds. i spent a year building it and roughly zero days telling anyone it existed. it died with no users.

i wrote the lesson down at the time: distribution isn't a phase that comes after building. it IS the building.

then i started vibedraft and ran a politer version of the same play. i was posting, sure. the account has 10,677 tweets on it. but posting isn't distributing, the same way having a product isn't having customers. nine out of ten of my posts earned zero bookmarks. nobody was saving anything, which means nobody was coming back, which means every post was a coin flip with no memory.

activity felt like distribution. the bookmark column says it wasn't.

mistake 2: i shipped features while the funnel had a hole in it

when growth stalls, building features feels like working. it's the most productive-looking way to avoid the real problem.

so vibedraft got better voice training, better scheduling, a better editor. meanwhile, the actual funnel looked like this: someone finds a post, maybe reads the blog, nods, and leaves. no email capture. no reason to return. the middle of the funnel wasn't leaky, it was missing.

a churned customer is feedback. a visitor you can never reach again is nothing. i optimized the product for the nine users i had instead of building the machine that finds the next ninety.

mistake 3: no number was allowed to hurt me

i knew the MRR the whole time. i just didn't publish it. private numbers are painless, and painless numbers don't change behavior.

nothing i shipped had a kill criterion either. no feature carried a "if this doesn't move X by date Y, it dies" note. so nothing ever officially failed, and i never got the signal to stop and redirect the hours.

that's why the counters on this site are live now. MRR from a verified listing, followers from the api, newsletter subscribers straight from the database. that subscriber counter says zero right now, and i left it public anyway. the number is allowed to hurt until it doesn't.

the checklist i should have been running

monthly. binary answers, no partial credit.

1. can a stranger hand me their email within 10 seconds of landing on the homepage?
2. is my revenue number public, somewhere i can't quietly edit it?
3. did the last 30 days of posts earn bookmarks from strangers, or just impressions?
4. does every live feature and surface have a written kill criterion with a date on it?
5. did distribution get as many hours as building this month?

four out of five or better, keep going. under that, stop building until it's fixed. this month i score 2 of 5. that's the whole reason this site exists.

$152 is tuition, not revenue

studentgrounds cost a year and returned zero users and one lesson. vibedraft cost three months of off-hours and returned $152, ten customers who came and went, and this list. the tuition is getting cheaper and the lessons are getting more specific. that's the trend i'm betting on.

vibedraft isn't dead, to be clear. it's live, it works, and i still use it every day. what died is the version of me that thought a good product markets itself.

if you want the next post-mortem while it's happening instead of after it's safe to tell, the letter goes out weekly. real numbers, win or loss.