I Ignored the Data and Launched Anyway. It Was the Right Call (Sometimes).

The data said no. The users said yes. I was stuck.

We were three months into Sonic Linker, and I had a decision to make. Our analytics showed that 80% of users never touched the advanced routing feature we were planning. Time on page? Low. Clicks? Even lower. If I ran this through any prioritization framework, it would rank dead last.

But then I'd get on a call with one of our power users, and they'd bring it up again. "When are you building custom routing logic? We need it." Not want. Need.

I shipped it. The data was right about one thing: most users didn't care. But the three who did? They became our best advocates. One of them brought in two referrals in the next month.

This is the tension every PM lives in. Data tells you what happened. Your gut tells you what might happen. And you have to decide which one to trust, knowing you'll be wrong sometimes either way.

Data works when the question is clear

Here's when I trust the numbers completely: when I'm optimizing something that already exists.

At Finvestfx, we had an onboarding flow that was losing 40% of users at step three. I didn't need intuition there. I needed a funnel chart and some A/B tests. We moved two form fields to a later step, and drop-off went down to 28% in two weeks.

That's a clean problem. The behavior exists, the baseline is clear, and you're just trying to improve it. Data is fantastic for this.

Same thing with pricing. When I helped price our forex reconciliation module, I didn't guess. I looked at usage patterns, ran cohort retention by plan type, and checked what percentage of users hit feature limits. The data told me exactly where to draw the line between tiers.

But data has a limitation: it can only tell you about the past. It can't tell you if something new will work. It can't tell you if a market is about to shift. And it definitely can't tell you if the three users asking for a feature represent a much larger silent group, or if they're just loud outliers.

Your gut works when you've earned it

I don't mean "trust your instincts" in some mystical way. I mean: if you've talked to 50 users in the last two months, your gut is actually pattern recognition. It's your brain synthesizing things the spreadsheet can't capture.

At Sonic Linker, I was on calls every week. I heard the same pain points in different words. I saw people's faces when they talked about their workflows. That's not soft skills fluff. That's context the data doesn't have.

When we were deciding whether to build an AI-powered link preview generator, the usage data was neutral. We didn't have enough history to know if people would use it. But I'd heard 15 people in the last month say some version of "I wish I didn't have to write these descriptions manually."

I trusted that pattern. We built it. It became one of our most-used features within six weeks.

The flip side: if you haven't talked to users recently, your gut is just bias. I learned this at NJ Group when I was coaching insurance advisors. I thought I knew what they needed because I'd been around the industry for a while. But when I actually sat in on their client calls, I realized I was completely wrong about their workflow. My gut was based on outdated assumptions, not current reality.

The real skill is knowing which mode you're in

Here's my actual decision framework now:

If the feature already exists and I'm trying to improve it: data wins. Run the test, check the metrics, optimize.

If I'm deciding whether to build something new and I've talked to fewer than 10 users about it in the last month: don't trust my gut. Go talk to more people first.

If I'm deciding whether to build something new, I've talked to a lot of users, and there's a consistent pattern even if it's not showing up in the dashboards yet: gut wins. But I set a clear success metric upfront so I can admit I was wrong later if needed.

If the data and my gut agree: ship fast, don't overthink it.

If they conflict: I ask myself, "Am I seeing something the data can't capture, or am I just annoyed that the data is telling me something I don't want to hear?"

What actually happened with that routing feature

Six months after I shipped the advanced routing logic at Sonic Linker, the data still looked bad. Only 12% of users touched it. But those 12% had a 6-month retention rate of 94%, compared to 68% overall.

Was I right to ignore the initial data? Maybe. Or maybe I got lucky. The honest answer is I'll never know for sure. But I do know this: if I'd only looked at the dashboard, I would have missed the signal entirely. And if I'd only listened to the loud users without checking anything after launch, I wouldn't have known whether it actually mattered.

The best PMs I know don't pick data or gut. They use both, they're honest about which one they're leaning on, and they're willing to be wrong. That's the real skill.