I almost shut down on the day we got our biggest break

Twelve months into building Atlas Vault — a small, opinionated treasury tool for early-stage startups — I had four paying customers, $1,650 in MRR, and savings down to under three months of runway.

I'd been working alone, sleeping six hours a night, and saying no to every consulting gig that came my way because I was afraid splitting my attention would kill the company. Splitting my attention being the only thing that would have actually paid the rent.

The problem wasn't product. Customers loved it. The problem was distribution. I'd written 38 blog posts, hosted 14 webinars, and shown up to every fintech meetup in two cities. None of it was scaling. I was the only growth channel, and I was running out.

Atlas Vault started after I watched my last company nearly mismanage its way into a cash crunch. We'd had $14M in the bank and almost no real-time visibility into where it was, who could move it, or what it was earning. The CFO managed everything in a spreadsheet that lived on her laptop.

I quit, took twelve weeks to build a v0, and launched it at a small fintech meetup. Three founders signed up that night. One of them is still my biggest customer.

For the first nine months, growth was steady. Two new customers a month, mostly from word of mouth and Twitter. Then it stalled. The same friend-of-friend network got saturated. Cold outbound didn't work — early-stage CFOs don't open emails from strangers. Content was a slow burn that wasn't burning fast enough.

I spent month eleven trying every distribution playbook I'd ever read. Nothing moved the needle.

The lonely part wasn't the work. It was the not-knowing. Was I one good week away from compounding, or three months away from the end? I genuinely could not tell.

I started having stomach pains in month ten. My partner — gently, repeatedly — asked me to consider going back to a job. I made every chart I could to convince him (and myself) that we were close. None of the charts were honest enough.

On a Tuesday in January I sat down at 11:30pm and wrote a wind-down email to my four customers. It said I was grateful, that I'd refund any prepaid annual plans, and that I'd open-source the tool so they could keep using it. I read it four times. I edited the subject line twice. At 11:47pm I had my finger on the send button.

Then Gmail dinged.

It was an inbound from a partner at a small accelerator I'd never pitched. They had 32 portfolio companies. Half of them needed exactly what I'd built. Could we set up a call?

I did not send the wind-down email. I scheduled the call for the next morning.

Within three weeks I'd onboarded eleven of their portfolio companies onto Atlas Vault at $290/month each. The accelerator partner became a paid advisor. Two of those eleven companies referred me to their friends at other accelerators within the next quarter.

MRR went from $1,650 to $4,840 in a month, then to $9,200 the month after. By month seven I was profitable. By month ten I'd hired my first contractor.

The accelerator inbound had not arrived because of any of my distribution efforts. It had arrived because one of my four original customers had told their accelerator partner, casually, in passing, that Atlas Vault was the thing they wished every portfolio company used.

1. 1
   The ground beneath you is firmer than your panic suggests.

   I had four customers. They loved the product. The math wasn't doomed; my distribution model was.

2. 2
   Word of mouth has a long tail.

   The customer conversation that saved my company happened months after the customer first signed up. Compound interest applies to delight, not just code.

3. 3
   Solo founders need a forcing function for honesty.

   I should have written that wind-down email three months earlier. The clarity it forced — about which customer relationships I was actually nurturing — would have made me act faster.

4. 4
   Channel partners beat content for early-stage B2B.

   One accelerator did more for me in three weeks than 38 blog posts did in twelve months.

5. 5
   Don't open-source the company until it's actually dead.

   I almost mistook a hard month for a dead company.