The hunch that cost me four months
I was absolutely certain that mid-market HR tech companies were drowning in competitor noise and desperately needed a better way to track what was being said about them. I had talked to three founders at a conference, all of them nodded vigorously, and I walked away feeling like I had cracked the code. Four months later, after building a feature set aimed squarely at that segment, I had five free trials and zero conversions. The nodding, it turned out, meant "interesting idea" not "I will pay for this."
Why founder intuition is both your best asset and your worst one
Every founder I respect runs on pattern recognition. You spot something broken before the market names it. You connect dots that analysts miss because they are looking at reports, not at the actual friction in people's days. That intuition is real and it is worth protecting.
But the same cognitive machinery that lets you see around corners also lets you build elaborate castles on a foundation of three data points and a good conference conversation. I have done this more than once. The HR tech detour was not even the most expensive version. I spent two months convinced that VC associates were the right entry point for selling investor research tools, because a partner at a fund told me his associates were overwhelmed. He was describing his problem. His associates were not the ones with budget or urgency to solve it.
What changed things for me was not becoming less intuitive. It was building a system that could argue back.
What I actually started doing differently
The first shift was separating signal-gathering from interpretation. When I was doing both in the same conversation, the interpretation always won. I would hear something ambiguous and immediately file it under the hypothesis I was already running. So I started recording discovery calls and reviewing them two days later, cold, without the emotional residue of the conversation. Things I had mentally labeled as strong confirmation looked very different on a reread. People were being polite. They were describing a past pain, not a current one. They were talking about what their company needed in theory, not what they personally had budget authority to buy.
The second shift was using social listening data as a check on my own assumptions before I built anything. This sounds obvious but I was not doing it systematically. I would run searches on competitor mentions, industry pain points, and the specific language people used when complaining publicly in forums, LinkedIn comments, and community threads. That language gap alone was instructive. I kept using the phrase "brand share of voice" in my positioning. Almost nobody in my actual target segment used that phrase. They were saying things like "I have no idea if our content is actually landing with the right people" and "our SDRs are going dark because they don't have good context before outreach." Different words, different frame, different product emphasis.
The third thing I did was build a small internal rubric before starting any new feature or segment push. Five questions. How many distinct people, unprompted, described this exact pain in the last 60 days. What is the current workaround they are using and are they paying for it already. Who specifically owns this problem inside the company and do they have discretionary budget. What would success look like to them in 90 days. And what is the second-order reason they might not buy even if they agree the problem is real. Running that rubric killed two features before I wasted engineering time on them. It felt wasteful to do that analysis. It was not.
The fourth shift was getting more honest about what the data from our own platform was telling me versus what I wanted it to say. We built an investor research component inside MentionFox, the investor suite, partly because I personally wanted that tool and assumed other operators and founders would too. Usage data told a more specific story. The people who actually leaned into it were not founders doing general market research. They were founders in active fundraising processes who needed fast competitive context before partner meetings. Totally different use case, totally different messaging, totally different urgency trigger. I had the right product and the wrong story about who needed it and when.
The fifth thing, and this is the uncomfortable one, is that I started writing down my hunches with a confidence score and a timestamp, then checking back on them quarterly. Most founders, including me, have selective memory about our predictions. We remember the ones that landed. The ones that did not tend to get quietly reclassified as "pivots" or "learnings." Keeping a literal log of what I predicted, how confident I was, and what actually happened has made me significantly more calibrated. My hit rate on strong-confidence hunches about enterprise segments is around 40 percent. That is not bad, but it means I should not be betting four months of engineering on any single one of them without external data as a check.
The practical version of this
If you are a founder and you are about to build something because you have a strong feeling, that feeling is data but it is not evidence. Before you scope the feature, run the rubric. Go find 20 to 30 public conversations - Reddit threads, LinkedIn posts, community Slack archives, whatever you can get - where people are describing the problem you think you are solving. If you cannot find those conversations, that is itself important information. Either the pain is real but private, which happens in some enterprise contexts, or the pain is not as acute as you believe.
Use whatever tools you have to do that listening systematically rather than sampling the conversations that already confirm what you think. The whole point is to find the evidence that argues against you. Your own brain will not surface that evidence voluntarily. It needs a process that forces the question.
Where MentionFox fits into this for me
I am not going to pretend objectivity here. I built parts of MentionFox specifically because I needed this discipline for myself and my team. The social listening layer is what I use to pressure-test positioning before we ship anything. The investor research tools inside the investor suite came directly from a real use case I kept running manually before we built it out. But the broader point holds regardless of what tools you use. The discipline matters more than the tool.
If you want to see how MentionFox handles the signal-gathering side of this - tracking mentions, surface-level sentiment, lead signals, and the AI visibility data that tells you how your brand is appearing in LLM-generated answers - the pricing page has the breakdown by use case and team size. No aggressive sales sequence, just the information.
My hunches are still useful. I just do not let them run unsupervised anymore.
If you found this useful, I write about solo-founder distribution, B2B SaaS, and what's actually working in the AI-search era over on my Substack (one post per week, no spam).
I'm building MentionFox - a B2B intelligence suite that combines brand mention tracking with AI-visibility (GEO) measurement, investor research, and outreach automation. There's a free tier and a 5-day trial of Pro at mentionfox.com/pricing.
Top comments (0)