I Made 3 Gut Calls at Sonic Linker That Saved Us Months. Data Would've Said No.

The numbers told me to optimize onboarding. I built a sharing feature instead.

When I joined Sonic Linker's founding team, we had a problem. Our AI platform was live, users were signing up, but activation was sitting at around 35%. The data was screaming: fix your onboarding flow. Drop-off was happening at step 2 of 4. Clear signal, right?

I spent two days staring at heatmaps and session recordings. Then I ignored all of it.

Instead, I pushed to build a team collaboration feature. Not because users were asking for it in droves. Not because we had survey data pointing to it. But because every single power user I talked to (all 12 of them) mentioned some version of the same pain: *"I wish I could just share this with my team without exporting and re-uploading."*

Two months later, that sharing feature became our wedge into B2B. Three enterprise pilots came directly from teams who started as individual users and wanted to bring colleagues in. The onboarding flow? Still not perfect. But it stopped mattering as much because we had a different growth motion.

Data would have optimized us into a slightly better version of what we already were. Gut took us somewhere new.

Here's when I trust data (and it's more often than you think)

I'm not anti-data. At Finvestfx, I lived in Mixpanel. When we were trying to improve client retention from 78% to 85%, I trusted the numbers completely. We ran cohort analyses, tracked feature usage across 20+ enterprise clients, and correlated activity patterns with churn risk.

The data told us that clients who adopted our automated reconciliation module in the first 30 days had a 92% retention rate. Those who didn't? 68%. So we changed our onboarding to front-load that feature. Retention hit 84% within two quarters.

That's a data decision. Why? Because the behavior already existed. I wasn't guessing what might work, I was scaling what already worked for a subset of users.

Here's my rule: If you're optimizing something that already exists, trust the data. If you're deciding what to build next, trust your gut (but make it an informed gut).

Data is great at telling you *what* is happening. It's terrible at telling you *why* it matters or *what* to do about it. That's where pattern recognition and judgment come in.

The "informed gut" isn't magic. It's just compressed experience.

At NJ Group, I was coaching 60+ insurance advisors and IFAs on product adoption. I had no analytics dashboard. No funnel metrics. Just conversations, feedback forms, and a lot of WhatsApp messages.

But I noticed something. Advisors who understood the *compliance angle* of our platform adopted it faster than those who only saw it as a CRM. The ones who got it weren't necessarily the tech-savvy ones. They were the ones who'd been burned by audits before.

I didn't have data to back this up. I had a hunch based on 15 conversations. So I changed our messaging in the next training session to lead with compliance and audit-readiness, not productivity. Adoption rate in that cohort jumped from around 40% to 65%.

That's gut. But it's not random. It's synthesized signal from qualitative inputs. The mistake people make is thinking gut decisions are uninformed. The best gut calls come from people who've talked to users obsessively, even if they haven't instrumented every click.

The real trick: know which game you're playing

Early stage (like Sonic Linker at month 3)? Gut wins. You don't have enough data to be statistically significant anyway, and you need to place bets on direction, not optimization.

Scaling stage (like Finvestfx with 20 clients and clear usage patterns)? Data wins. You have real behavior to learn from, and your job is to do more of what works.

The trap is using the wrong tool for the stage you're in. I've seen founders at 100 users obsess over conversion rate improvements that move the needle by 2%. I've also seen PMs at scale ignore obvious churn signals because they're too busy chasing their "vision."

At Sonic Linker, we shipped our core product in 3 months because we made fast gut calls on what to build. At Finvestfx, I spent 6 months improving retention by analyzing the hell out of client behavior data.

Both were right. Just different games.

What I actually do now

I don't make gut calls in a vacuum. I stack qualitative signal until a pattern emerges, then I move. If I've heard the same pain point from 5+ users in different contexts, that's enough. I'm not waiting for a statistically significant sample.

But once something is live, I instrument it properly and let the data tell me if I was right. Gut gets you to the bet. Data tells you if you won.

The worst thing you can do? Pretend you're being "data-driven" when you're just cherry-picking numbers that confirm what you already wanted to build. I've done that. It's gut with extra steps and a fake air of rigor.

If you're going to trust your gut, own it. If you're going to trust data, actually change your mind when it disagrees with you.