I wrote my own resignation, then my cofounder rewrote our entire pricing model

By the end of year two, Mainline had 11,000 free users, 380 paying customers at $19/month, and a churn rate that quietly ate 7% of MRR every month. We were a creator-tools company in a category where every competitor priced like Spotify.

The math didn't work. Our gross margin was 31% — destroyed by AI inference costs we hadn't modeled when we set the original price. Our blended CAC was $190. Our LTV, after churn, was $214. We were paying ourselves to acquire customers.

Every board update we showed the same chart with a slightly different optimistic projection. My cofounder Pierre and I both knew it wasn't going to bend.

Then our largest single customer — a podcast network paying $3,200/month on a custom plan — churned without a meeting.

Pierre and I had met at a creator-economy company that exited well. We left with the conviction that the next layer of tools would be AI-powered and built specifically for serious independent operators, not hobbyists.

The v1 launch went better than expected. Product Hunt #1. Twitter virality. 4,000 signups in a week. We'd priced at $19/month because everyone in our category did.

That number followed us for two years. We'd debate it on every leadership offsite. We'd sketch new tiers on whiteboards. We'd never ship them, because the prevailing fear was always the same: "Our existing users will revolt. Our pipeline will collapse. We'll lose press."

So we kept shipping features instead. Each new feature added cost. None of them moved the price.

The week the podcast network churned, I stopped sleeping. I'd lie awake doing the unit economics in my head, the way other people count sheep.

On a Tuesday morning I opened a Google Doc titled "Resignation — draft." I wrote two paragraphs. I cried while writing them. Pierre saw the doc in our shared drive (I'd forgotten about default sharing) and called me thirty minutes later.

We met at a coffee shop in the Mission. He asked one question: "If we pretended our existing pricing didn't exist, what would we charge?" I said $190 a month for the bottom plan. He said $290. We argued about it for two hours and ended up at $249 with a $79 hobbyist tier we'd never advertise.

We spent Wednesday and Thursday in his apartment rewriting the pricing page, the onboarding, and the upgrade flow. We didn't tell the team until Friday morning.

The grandfather email was the hardest thing I've ever written. Existing customers kept their $19 price for life. New signups went straight to $79 / $249. We were braced for a Reddit thread.

The Reddit thread never came.

In week one we lost three trial signups who would have converted at the old price. In week two, we closed seven new logos at $249. By the end of month one, new MRR had grown 4.2x even with the lower volume of paid conversions, because each new customer was worth ten times more.

Gross margin moved from 31% to 64% within ninety days. We rebuilt the funnel around a much narrower ICP — agencies and small in-house creator teams, not solo hobbyists. CAC dropped because we stopped paying for top-of-funnel traffic that would never afford the new price. The first profitable month came eight weeks after the change.

I never sent the resignation.

1. 1
   Pricing is positioning. Positioning is the company.

   When we changed the number, we changed the customers, the support load, the roadmap, and the meetings on our calendars.

2. 2
   The "user revolt" is almost always a fantasy.

   Real users sort themselves into the tier they can afford. The loud ones on Twitter are rarely the ones paying you.

3. 3
   Grandfathering buys you the right to be brave.

   Promising existing customers their price for life is cheap insurance against the worst-case backlash.

4. 4
   A resignation draft is data.

   The fact that one of us was ready to walk meant something structural was wrong. We should have run the pricing experiment a year earlier.

5. 5
   If your unit economics require optimism, your pricing is wrong.

   Spreadsheets that only work in the best case are not models — they're prayers.