Why we killed our SaaS at $12K MRR (a post-mortem)

There's a particular kind of startup death that doesn't look like death. Revenue is up and to the right. The graph you'd put in an investor update looks great. And the two founders quietly know it's over.

This is the post-mortem of a B2B scheduling tool — we'll call it the product "Cadence" at the founders' request — that reached $12,400 in MRR with about 140 paying accounts and then shut down four months later. The co-founder who walked us through it, told as told to Joaquín del Río, asked that we lead with the lesson and not the logo. So: they shut down a profitable-on-paper product on purpose, and they'd do it again.

The graph that lied

Cadence launched in early 2024 and found early traction with small agencies. By mid-2025 it was at $12.4K MRR, growing roughly 8% month over month, with two founders and one part-time contractor. On a slide, that's a company.

"Every metric on the dashboard was green. The only red number was the one we weren't tracking."

The number they weren't tracking carefully enough was net revenue retention. New logos kept the top line growing. Underneath, accounts were leaving almost as fast as they arrived. Gross churn was running near 6% monthly — survivable if expansion offsets it, fatal if it doesn't. Cadence had almost no expansion. Nobody upgraded. They just left when the pain they'd hired Cadence to solve went away or got worse.

The unit economics, unflattering edition

Here's what they were willing to share, with the caveat that these are their internal figures and were never audited.

• MRR at peak: $12,400
• Paying accounts: ~140, average revenue per account ~$88/month
• Blended CAC: ~$310 (mostly founder time valued at a modest rate, plus ad experiments)
• Average account lifetime: ~9 months
• Lifetime value: ~$790 gross, before support costs

On paper LTV:CAC of roughly 2.5:1 looks fine. The problem is what that average hides. The agencies — their best customers — stayed 14+ months. A flood of solo users they acquired through a viral Product Hunt week stayed under four months and generated the bulk of support tickets.

"We optimized for the customers who were easiest to acquire and hardest to keep. It took us a year to notice we'd built the wrong funnel."

The fork they didn't take in time

In the autumn of 2025 they had a real decision. Option A: go upmarket, rebuild for the agencies, raise prices, accept slower growth and a smaller, stickier base. Option B: keep riding the solo-user wave that made the graph look good.

They chose B for three months without admitting they were choosing. "We didn't decide to chase the wrong customer," the co-founder says. "We just kept doing the thing that produced the dopamine of a growing number."

By the time they committed to the agency pivot, two competitors had moved into that segment with more capital, and the founders were tired in a way that's hard to put in a deck.

The cost nobody models

The financial post-mortem is tidy. The human one isn't. Both founders had taken 18 months of below-market pay. One had moved cities for the company. The product was making enough to cover its own server costs (~$900/month) and a little more, but not enough to pay two adults a real salary in a market where they could each earn well.

"The trap of $12K MRR," the co-founder says, "is that it's too much to walk away from casually and too little to live on. It's the most dangerous number. Zero is clarifying. $50K is a business. $12K is a guilt machine."

"Zero is clarifying. Fifty grand is a business. Twelve grand is a guilt machine."

How they actually shut it down

This is the part they're proud of, and it's why they let us publish.

They gave customers 90 days' notice, longer than any contract required. They wrote a migration guide to two competitors, including the one they'd lost the segment to. They open-sourced the parts of the codebase that weren't sensitive. And they refunded the most recent annual payments in full, which cost them about $4,200 they didn't have to spend.

"We wanted the shutdown to be the most honest thing we did," the co-founder says. "The product was mediocre. The exit didn't have to be."

Three former customers have since become customers of the founders' next thing. That's not a coincidence. People remember how you end.

What they'd do differently

The lessons, in their words, lightly edited.

Track retention before you track growth. "If we'd put NRR on the main dashboard in month three, we'd have made the agency call a year earlier."

Pick the customer you want to keep, not the one who's easy to get. "A Product Hunt spike is not a market. It's a sugar high with a support cost."

Name the fork out loud. "Indecision is a decision that compounds. We lost three months to not saying the obvious sentence to each other."

Model the founders' salaries as a real cost. "A business that can't pay you is a job that pays in equity you're printing yourself."

The number after the number

Cadence returned its last server invoice to zero in February 2026. Both founders are now building separately — one in developer tooling, one back in agency-land, this time as the customer rather than the vendor.

"I don't think of it as a failure anymore," the co-founder says. "I think of it as the most expensive, most useful course I ever took. Tuition was eighteen months and a graph that lied to me."

The dashboard they should have built

We asked the co-founder to design, in hindsight, the dashboard that would have saved Cadence. The answer is instructive because it's so much simpler than what they actually watched.

"We had a beautiful analytics setup," they say. "Funnels, cohort charts, twelve kinds of engagement metric. And the one screen we needed had three numbers on it." Those three: net revenue retention, logo retention by segment, and months of runway at current burn.

"NRR would have told us the base was leaking. Logo retention by segment would have screamed that agencies stayed and solo users didn't — we had that data, we just never split it. And runway would have told us how many months we had to act before the decision made itself." Instead, the dashboard they checked every morning led with signups and MRR, "the two numbers most likely to make you feel good and least likely to tell you the truth."

"We had twelve metrics and needed three. The other nine were there to make us feel busy."

The deeper lesson, they say, is that dashboards encode what you're afraid of. "We built a dashboard that celebrated growth because growth was the story we wanted. A dashboard that led with retention would have forced a conversation we were avoiding. The metrics you choose are a confession."

What $12K MRR did to the partnership

The part founders rarely publish is what the slow death did to the relationship between the two people living it. The co-founder was unusually willing to talk about it.

"For about six months we stopped being honest with each other, and we didn't notice," they say. "Not lying — just both quietly protecting the other from the conclusion we'd each privately reached." One founder thought the solo-user strategy was a dead end by autumn; the other suspected the same but kept building features because shipping felt better than confronting it. "We were each waiting for the other to say the hard thing. That's a failure mode of two reasonable people who like each other."

The thing that broke the silence was mundane: a quarterly numbers review where the runway figure was simply too stark to talk around. "Spreadsheets are good for that. A number on a screen says the sentence neither of you wants to say out loud."

They're still close, which they attribute almost entirely to the shutdown. "If we'd let it grind us into resentment, we'd have lost the company and the friendship. Ending it cleanly cost us a product and saved us a relationship. Cheap, in the end."

The advice they actually give now

When other founders ask, the co-founder has stopped giving tactical advice and gives one structural piece instead: decide, in advance, what number would make you stop.

"Before you're emotionally invested, write down the condition that means this isn't working — a churn rate, a runway threshold, a date. Put it somewhere you'll see it. Because in the moment, you will always find a reason to keep going. The you who hasn't fallen in love yet is the only one who can set an honest stop-loss."

"Set the stop-loss before you fall in love. The founder in love can't see the exit."

It's the advice they wish someone had pushed on them at $5K MRR, "when it was still a hobby and not yet a guilt machine."

The number that still stings

We asked the co-founder which figure, of all of them, they still think about. It wasn't the MRR or the runway. It was lifetime months by segment — the gap between the agencies who stayed 14+ months and the solo users who left in under four.

"That number existed the whole time," they say. "It was sitting in our database from month four. If I'd run that one query and put it on the wall, I'd have seen the entire future of the company in a single bar chart. The agencies were a business. The solo users were a treadmill. And we sprinted on the treadmill for a year because it produced motion."

The reason they didn't run the query, they admit, is the most human part of the whole story. "Some part of me didn't want to know. The solo-user growth felt good. Running the segmented retention query risked replacing a number that made me happy with one that demanded a hard decision. So I didn't run it. That's not an analytics failure. That's a courage failure dressed up as an oversight."

"I didn't run the query that would've saved us, because some part of me didn't want the answer. That's not a data problem. That's a courage problem."

What they took from it, into the next company, is a discipline: once a quarter, deliberately run the query you're most afraid of. "The metric you're avoiding is the metric that's about to matter. I learned that for the price of a company."

There's a second discipline, too, born from the same regret. In the new venture, the co-founder writes a short "red team" memo every quarter — a single page arguing, as persuasively as they can, that the company is actually failing. "You force yourself to make the bear case in your own voice. It's uncomfortable, which is exactly why it works. If the strongest argument against you is weak, that's real signal. If it's strong and you can't answer it, you just found your next quarter's priority before it found you." At Cadence, they never once sat down and tried to argue against themselves. "We were too busy celebrating a graph. The bear case was writing itself in the database and we weren't reading it."