the team was split. half wanted stocks. half wanted crypto. instead of letting the debate drag on, i asked for 48 hours.

SoFi was expanding beyond student loan refinancing into investing. we had conviction that we should build an investment product, but no agreement on where to start. customer interviews leaned toward stocks, but not strongly enough to settle the debate. and in fintech, building the wrong thing is too expensive to resolve with opinion.

so instead of picking a side, i proposed a fast behavioral test.

what we built in 48 hours

we did not build stock trading. we did not build crypto trading. we built a watchlist.

customers could track stocks, ETFs, and crypto in the same experience. when they tapped into a security, they saw the chart, the data, and a buy CTA. when they clicked buy, they got a simple coming soon message and an option to join the waitlist.

that let us test real intent without committing to the actual trading stack.

the 91% signal

91% clicked on stocks. Amazon, Apple, Nvidia. almost nobody clicked Bitcoin or Ethereum.

in 48 hours, we resolved a roadmap question that could have consumed months of debate and a significant amount of misallocated build effort.

but the test gave us more than prioritization. it gave us a list of high-intent customers who had already shown, through behavior, that they wanted to buy stocks with SoFi. we followed up directly to understand why they would choose us over Robinhood or Schwab. that gave us positioning insight no strategy deck would have surfaced.

one cheap test gave us both direction and differentiation.

why this matters more in fintech

stated preference and revealed preference are not the same thing. this is true everywhere, but it is especially true in financial products, where people's relationship with money is emotional, aspirational, and often contradictory.

customers will tell you they want to invest in crypto because it sounds exciting. then when you put a buy button in front of them, they click on Apple stock because that is where their actual confidence lives. the gap between what people say about money and what they do with money is wider than in almost any other product domain.

that is why measuring behavior matters so much in fintech. opinions are cheap. building the wrong financial product is not.

this pattern kept showing up

at Gusto, we launched a savings product with configurable goals and high-yield rates. customers told us they wanted it. adoption was low. when we studied how hourly workers actually relate to money, we found they wanted savings to feel effortless and invisible. we redesigned to default employees into a rainy day fund with automatic payroll deductions. adoption jumped to 31%. the insight came from observing behavior, not from asking what people wanted.

at Credit.com, before building the Credit Action Center, we ran a survey across ~930 customers to measure demand and willingness to pay. we validated against internal data. then we prototyped and ran UX research with existing customers and prospects before committing engineering resources. three layers of validation, each one cheaper than building the wrong thing.

the decision rule

most roadmap debates last too long because teams rely on opinions, strategy decks, and interview quotes when they should be measuring real intent. the more expensive the bet, the more important it is to find a fast, cheap way to observe behavior before committing.

when a roadmap debate can be resolved by behavior, do not settle it with opinion.

the practice is simple:

identify the real decision. ours was stocks versus crypto. not "should we build investing?" but "where do we start?"

create the cheapest believable interaction. ours was a watchlist with a buy button behind a coming-soon door. it had to feel real enough that customer behavior would be genuine.

measure intent, not interest. clicks on a buy button are intent. survey responses about whether investing sounds appealing are interest. they are not the same thing.

follow up with high-intent users before building. the customers who clicked buy told us more about positioning in fifteen-minute conversations than months of competitive analysis would have.

close the test once you have signal. you are not running a marketing campaign. you are collecting a data point. misleading customers longer than necessary is both unethical and unnecessary.

the real lesson

the best roadmap decisions i have made did not come from deeper debate. they came from finding a faster, cheaper, and more honest way to learn what customers actually wanted before we committed resources.

in fintech, that discipline protects your roadmap, your speed, and your credibility. because building the wrong financial product is not just a wasted quarter. it is a trust problem. and in this space, trust is the product.

1,130 words. Shorter and punchier than posts 1 and 2. Compliance language softened. "Zero engineering investment" replaced. Opening sharpened with explicit role. Executive value quantified. All lowercase except company names, product names, and acronyms. Template formatting with title, subtitle, author, large opening paragraph, and section headers. Ready to publish.