Why Startups Fail for People Reasons, and Why Most of It Is Predictable

I watched my own startup's team fall apart.

Not because we didn't get along — we did. But we had fundamentally different operating modes. I was wired to experiment, move fast, break things, and learn from what broke. My cofounder needed structure, certainty, and proof before acting. Neither approach was wrong. But together, under pressure, they created friction that compounded quietly until the partnership couldn't hold.

Looking back, every sign was there before we started. We just didn't have a framework to see it.

That experience is a big part of why I built Tellstone. But the more I talk to venture studios, accelerators, and early-stage operators, the more I realize my story isn't unusual. It's the norm. The details change, but the pattern doesn't.

The Problem Everyone Knows But Nobody Measures

You've probably seen the stat. Sixty-five percent of high-potential startups fail because of people issues rather than product or market issues. It's usually credited to Noam Wasserman's work at Harvard, but the number traces back to a 1989 survey of 49 venture capitalists about 96 companies, and it's been repeated so often that almost nobody checks it anymore.

The better-sourced version of the same point is stronger anyway. In a survey of 885 investors, 55% named the team as the single most important factor in their failed investments, and 95% called it an important factor, ranking it above product, market and timing (we lay out the research here). Not a reason among several. The reason.

But here's what's strange. Everyone knows this. Every studio operator, every accelerator director, every VC — they all nod when you bring it up. And then almost nobody does anything structured about it before committing capital and resources to a team.

I had a conversation recently with a managing director at an accelerator. He was helping a founder evaluate a potential cofounder. You know what his process looked like? He had them both take a Gallup assessment, then he threw the results into ChatGPT to see if they'd work well together.

He's not an outlier. That's what most of the market looks like right now. People are already trying to solve this problem — they're just doing it with tools that weren't built for it, held together with duct tape and gut instinct.

I reached out to a managing director who's spent years building and investing in early-stage teams. I asked him what risks he thought were actually predictable. Without hesitating, he rattled off a list: misaligned expectations, role confusion, work ethic mismatches, personal life changes midstream, gaps in actual risk appetite, and bad interpersonal dynamics that show up under stress.

He didn't need to think about it. He'd lived every one of those. Every operator has this list somewhere in their head. They've seen these patterns destroy teams. They just don't have a systematic way to catch them before the commitment is made.

Why Most of This Risk Is Predictable

There's a school of thought that says you can't predict any of this. That team dynamics are too messy, too human, too dependent on context to ever model. I spoke with a studio founder who's started over a thousand companies, and even he acknowledged that perspective exists.

But I don't buy it. And neither does he.

Here's what I've learned after three years of building assessments and analyzing teams: team risk is not ephemeral. Humans are pre-wired. When you put certain personality combinations together, real problem patterns emerge — not randomly, but predictably.

Some of it is about friction between people. A founder who needs to move fast paired with a cofounder who needs consensus before every decision. That's not a mystery. That's a predictable collision that you can see in the data before it ever shows up in a missed deadline or a blown-up partnership.

But some of it isn't about conflict at all. It's about gaps. A founding team of three deep, introverted thinkers might build beautiful technology and never close a single customer. That's not a personality clash — it's a missing piece. And it's just as predictable if you're looking at the right data.

We've analyzed roughly 30 teams so far. In the cases where we've been able to confirm whether our flagged risks were real — where we could go back and check — we've been right 8 out of 9 times.

That's early data. I'm not pretending it's a peer-reviewed study. But being right 8 of 9 on a small set tells me something real is working, especially when the people closest to those teams keep telling us the same thing: "Yeah, that's exactly right."

One example that sticks with me. We flagged a time management bottleneck in a CTO who'd been working with his cofounder for years. This wasn't a new team — they had a long history together. When the CEO saw the result, his reaction wasn't surprise. It was relief. He told me he'd been managing around that exact issue for eight years. The risk was real, but it was hidden — the team had unconsciously built systems to mask it.

That's the kind of thing interviews will never catch.

What Studios Are Doing Instead (and Why It Falls Short)

Most venture studios and accelerators that do anything structured at all are using tools like DISC, Predictive Index, Gallup StrengthsFinder, or Working Genius. These aren't bad tools. They've been around for decades and they measure what they're designed to measure.

But they weren't designed for this.

They're episodic — you take the assessment when something is at stake, get a profile, maybe have a conversation about it, and then it goes away. They measure individuals in isolation. They don't tell you what happens when these specific people try to build this specific thing together, under these specific conditions.

A PE firm called Vista Equity Partners requires every potential portfolio company executive to take a psychometric assessment before they'll close a deal. They've walked away from transactions they loved — great company, great market, great financials — because the founder's assessment results didn't clear their bar. That's how seriously the highest-stakes investors take this.

But venture studios, who are arguably making even more people-dependent bets with less margin for error, are doing it with tools that were built for corporate HR departments. There's a gap here.

What Structured Team Diligence Actually Looks Like

Studios already do rigorous diligence on markets, technology, and financials before making a bet. Team diligence is the same rigor, applied to the people.

It means assessing founders — not just their resumes, but how they actually work, communicate, make decisions, and handle stress. It means simulating team scenarios before committing to them, so you can see where the friction and the gaps are before they cost you six months and $300K. It means identifying what's missing from a team — not just "we need a technical cofounder," but specifically what work style, decision-making profile, and soft skills this particular team is missing.

And critically, it means tracking whether your predictions were right. Did the risk you flagged actually show up? Did the hire you recommended actually work out? Because that feedback loop is what turns a one-time assessment into a compounding dataset that gets smarter every time you use it.

That's the part that changes everything. A studio that tracks outcomes across 10, 20, 50 team decisions starts to build something no one else has — a proprietary model of what actually works at their studio, calibrated to their track record. Not generic patterns from a research paper. Real signal from real decisions.

We published a case study recently with Purdue Innovates, where Tellstone was used during final accelerator cohort selection. The program manager said it helped them validate instincts they already had, surface traits they didn't see in interviews, and identify where teams might need more support. You can read the full case study here.

The Bottom Line

Team risk is measurable. It's predictable. And it's addressable — if you're willing to look at the data instead of relying on gut feel alone.

That failure rate isn't a law of nature. It's a reflection of how little structured attention gets paid to the most important variable in early-stage ventures, which is the people.

The studios that start measuring this now — that build a real dataset of team decisions and outcomes — will have a compounding advantage that gets harder to replicate with every decision they make. The ones that keep relying on vibe checks will keep learning the hard way.

I know because I learned the hard way myself.