--- title: "The capital allocation mistake that looks like success until the market goes quiet" url: "https://cfodrive.com/insight/the-capital-allocation-mistake-that-looks-like-success-until-the-market-goes-quiet/" author: "Nick Sawinyh" published: "2026-10-01" updated: "2026-10-01" --- # The capital allocation mistake that looks like success until the market goes quiet I've made one financial decision that I'd call a genuine error rather than bad luck, and for two years it looked like the best decision I'd ever made. The way it hid is the useful part, so I want to describe it precisely. It's now the first filter I put on any new bet. In 2020 I co-founded an analytics platform for decentralized exchanges. The sector was heating up fast. We raised $6M across two rounds, the product hit 100K daily users at peak, and it processed over a billion dollars in cumulative volume. By any dashboard a finance team would build, that was demand, and we allocated capital accordingly: more product depth, more people. The number I didn't have on the dashboard was how many of those users would still be there when the sector stopped being interesting. When it did, the answer turned out to be a fraction of what the traffic had implied. The durable customers were a small group with a specific need, and I'd built for the crowd instead of for them. ### Attention and demand produce the same chart Attention is people engaging with a thing because it's the thing people are engaging with: press, social volume, inbound interest, usage that tracks a sector's temperature. Demand is people paying for something, repeatedly, when nobody's watching and nothing is trending. For anyone allocating capital, the trouble is that for a long stretch these two produce identical charts. Users are up and revenue is up, and retention looks fine because retention during a hot period is retention of people who haven't yet had a reason to leave. Every leading indicator confirms the story. The only number that separates the two arrives late, when you find out what the customer does once the sector goes quiet, and by then the capital is spent. So the standard approach, forecast on trailing usage and fund the forecast, is blind to the one distinction that matters most. The forecast measures the wrong quantity and reports it with confidence. ### The three questions I now ask, in order I don't think a model resolves this, and I'd distrust anyone selling one. What I have instead is a short ordered set of questions that I put ahead of any spreadsheet. Who is paying today? Interest is free and abundant during a hot period. Payment is a specific act by a specific person. If nobody is paying yet, the opportunity is a bet, and it should be sized and described as a bet instead of carried on the books as a business with early revenue. What does the customer do when the sector goes quiet? If the honest answer is "they stop," the revenue is attention. Sometimes you can test this early: find the customers who'd use the product if the sector were boring, and look at whether their usage pattern differs from the crowd's. I had the data to do that and never ran the query. What's the cost of being wrong, and is it reversible? This is the one that changes decisions. I'll fund a cheap, reversible bet on attention without much analysis, because being wrong costs little and being right compounds. I won't fund an expensive, irreversible one, whatever the dashboard says. Headcount and product depth that only make sense at scale are irreversible, and they're exactly what a hot market tempts you to buy. The third question only means something once you've answered the first two honestly, which is why the order matters. ### The counter-example that kept me honest I've also run the opposite experiment, though not on purpose. Since 2019 I've operated a niche media site in the same broad sector. It has never bought a link, never run a paid campaign, and never raised money. It grew to a 55K+ audience on organic traffic alone, and it came through several algorithm updates that flattened larger, better-funded competitors. Its costs are small and its revenue is unglamorous, and it has outlasted the venture-backed company by a wide margin. The venture-backed company did things the media site never could, and if the goal was to capture a window, it captured it. But the site is the cleanest demonstration I have of the difference. Its audience was there for a reason that had nothing to do with the sector's temperature, so it never had a hot-market cohort to lose. From the inside, demand is boring. ### What this means for risk analysis Most risk frameworks I've seen treat market risk as a distribution around a forecast: the forecast is the expected case, and risk is the variance. That model assumes the forecast is measuring something real. The attention-versus-demand problem is closer to a measurement error than a variance, and it doesn't show up in the distribution at all, because every scenario in the model shares the same wrong assumption about what the users are. The practical adjustment is to build one more scenario that doesn't come from the model: the sector goes cold, attention-driven usage goes to zero, and the only revenue left is from customers who'd have paid regardless. If you can't estimate that number, that's the finding. It means you don't know which of the two you're funding, and every irreversible commitment should wait until you do. ### What I'm not sure of This framework, applied in 2020, would have told me to build smaller, and smaller might have meant missing the window entirely. Capturing a hot market is a legitimate strategy, and some companies have captured theirs and converted the attention into durable demand before it faded. I'm not certain the cautious version wins in every case. I am certain that I didn't know which of the two I was buying, and the not knowing was the mistake. If you allocate capital and you can't say, for your biggest growth line, what fraction of it survives the market going quiet, that's the number to go find before the next board meeting. Finding it the way I did costs more. --- Nick Sawinyh is the founder of DeFiPrime, an independent media and research site covering decentralized finance since 2019. He previously co-founded and led a venture-backed DEX analytics platform and has spent over a decade taking technically complex products to market.