We shut down closedwon at the end of 2025. After working at various startups before, it was my first attempt to build one, I could call my own. I want to write down what I've learned while the memories are still sharp.

I built it with Vignesh. He was once my boss, colleague at other times, and a friend through all of it.

Closedwon came out of a problem we had experienced in common. Pricing is hard to explain. If you sell SaaS on usage-based pricing, it is further pronounced. SaaS pricing is not easy to model and CFOs don't sign off on numbers they can't pin down.

Our early interviews with sales execs reinforced this. We heard deals that should close in a week take a quarter, or a lot of them never close at all.

Our first version, a barebones app let sellers create a quote with standard steps, but let buyers simulate their usage in real time. They could fine tune usage dials, and see your estimated spend in seconds. We built support for complex pricing models, such as ramp pricing and tiered models. These were hard enough to even model on spreadsheets. The early demos landed well with sales and RevOps leaders. For a while it felt like we were onto something. In the months ahead, we went on to build a full dealroom, agentic quote creation from call transcripts and buyer intelligence.

We built it on Bubble. No engineers on payroll and no infra to babysit. We often went from new idea to a first demo in a week. Speed and low cost was our edge. I believed we were onto something meaningful.

To my disappointment, it still didn't work as a business.

1. Vitamin, not a painkiller

Every call went well. Sales ops people liked the demo but we didn't get meaningful champions for our product. You can have a product that solves a problem and still lose. The person who feels the pain is not the person who signs the check. I had read about this. Reading about it and watching it happen to your own deals are not the same thing.

The honest version of our situation is this. We were a nice-to-have in a year when nobody was buying nice-to-haves. Could we have reframed it? Maybe. Priced it as ROI per closed deal? Maybe. But the reframing needed to happen before we built the thing, not after.

2. Moving fast is an arbitrage, not a moat

We built Closedwon on Bubble. In mid-2024 that felt like cheating. Everyone else was wrangling services and frameworks. We were shipping features in a week. Speed was the whole thesis. If we got a few clients, we were going to out-ship them.

By mid-2025 this advantage had all disappeared, and suddenly. Cursor and Claude Code got good enough that a solo engineer writing real code could move as fast as we did on Bubble. Often faster. The thing we had traded flexibility for, which was speed, stopped being scarce.

Moats made of distribution, data, or hard-won customer trust tend to last. But moats built on tools can vanish overnight as tech evolves. We knew this in the abstract. We did not apply it to ourselves. When you are the one moving fast, it feels like a skill. It is usually just a temporary arbitrage.

Knowing when to stop

This was hard. We were tempted to add more features, and for a while we did, but we had to draw a line.

Startups run on belief. Belief has a way of surviving evidence. I could tell you a story where we pivoted, found the right ICP, raised a round, and it worked. Founders tell themselves that story every day. Sometimes they are right. We had to be honest about whether we were in that phase, or whether we were going to spend another year forcing a fit the market kept politely declining.

We chose to stop. I still do not know for certain it was the right call. I think it was. The version of us that kept pushing would have learned less than the version that steps back.

Closedwon is done. I am glad we built it.