Two weeks ago I asked my developer how long it would take to add a new feature to the Coachstack portal.
The answer came back: about three months.
That night I sat down with Claude Code and scoped the same feature myself. My estimate, based on what I've actually been shipping lately: two or three days.
I stared at those two numbers for a long time. Because I knew what they meant, and I didn't want them to mean it.
What I'm walking away from
For the last year and a half, Coachstack has lived in Bubble, a no-code platform. A no-code developer built the first MVP in three weeks. I tested it, ran it through a pilot with real coaches last year, got genuinely good feedback, and kept building on top of it.
I have poured thousands of hours into that app. Tens of thousands of dollars. I sold a car to fund the early days of this company, and a lot of that money went into this exact product.
And at the end of this month, I'm sunsetting it.
The trap I almost fell into
Here's the thing about sunk costs. Everyone knows the textbook definition. I have an MBA. I could have explained the sunk cost fallacy to you years ago over a beer, confidently, like someone who would obviously never fall for it.
Then it was my thousands of hours. My tens of thousands of dollars. My eighteen months.
When it's yours, the math stops feeling like math. Every argument for staying the course sounds reasonable. We've come so far. The pilot went well. Switching now would mean starting over. You can keep a dying decision alive for a very long time on those sentences.
The question that finally cut through wasn't "how much have I invested?" It was the opposite question:
If I were starting Coachstack today, knowing everything I know now, would I build it this way?
The answer was no. Not even close. And once I admitted that, keeping the Bubble app alive wasn't loyalty or grit. It was just paying interest on a decision I'd already outgrown.
How I got here
Over the last six months, something changed in how I work. As the new AI models kept getting better, I got deeper and deeper into Claude Code. I learned to set up loops, automate development tasks, and build features in days that used to take me months to spec and hand off.
The strange part is that I didn't set out to rebuild Coachstack at all.
It started with me building the website for Solo Founder Coach, my coaching practice. Just riffing. Building the brand, the logo, the pages. Then I wanted a newsletter, and instead of picking between Beehiiv and Substack and Ghost, I caught myself thinking: there are pieces of each one I like. What if I just built my own?
Then I thought about who would actually write the newsletter. I'd been paying a ghostwriter, who is excellent, nearly $2,000 a month, and I couldn't justify that anymore. Writing it entirely myself would take five-plus hours a week I don't have. So I built a newsletter wizard that uses the same process world-class ghostwriters use: a real content strategy, an ideal reader profile, content pillars, a tone of voice with dos and don'ts, a proper content brief. It now drafts a full newsletter in about thirty seconds, and it's genuinely good.
Then it repurposes that newsletter into a handful of social posts, hub-and-spoke style. One focused hour of marketing instead of ten scattered ones.
None of that was a master plan. I just kept following the thing that was working. And at some point I looked up and realized I had accidentally built the foundation of the new Coachstack.
Building it backwards
Here's what's funny. The new build is almost the exact opposite of how I approached the Bubble app.
Back then, I believed I had to build the full foundation first. Client portal, billing, scheduling, contracts, video, session notes. The whole house, before anyone moved in.
This time I'm starting where the pain actually is. After 30+ coach interviews and hundreds of hours of research, the loudest problem was never "I need a better client portal." It was "I can't get enough clients." Lead generation. Marketing. Sales.
So the new Coachstack starts there: the coach's website, their newsletter, their LinkedIn content, their lead magnet. The growth engine first. The back office later, if and when coaches tell me they truly need it.
It took me a year and a half and a painful goodbye to learn that order matters.
The part that still stings
I want to be honest about something, because that's the deal with this newsletter.
This decision hurt. The developer I worked with is talented, and that first three-week MVP is the reason Coachstack exists at all. The pilot coaches gave me feedback that shaped everything I'm building now. None of that was wasted, even though the code is being retired.
But I'd be lying if I said I didn't grieve it a little. You don't spend eighteen months building something and then turn it off without feeling it. There was no one in the room to talk me into it or out of it. Just me, two estimates, and a long Friday of sitting with what they meant.
The takeaway, if you want one
If you're holding onto something right now, a tool, a product, a strategy, a vendor, maybe even a whole business model, try asking the forward-looking question instead of the backward-looking one:
Not "how much have I put into this?"
But "knowing what I know today, would I choose this again?"
If the answer is no, the money and the hours are already spent either way. The only thing still on the table is your future.
I chose mine this week. It cost me eighteen months of work to do it, and I think it's the best decision I've made since starting Coachstack.
See you next Saturday.
Peter
