Why VCs fund "crazy" not good companies
Ben Lerer of Lerer Hippeau on conviction over consensus, the Peloton miss, and why backing sensible businesses is a mistake.
The GTMnow podcast recently sat down with Ben Lerer, Managing Partner of Lerer Hippeau, a firm that has deployed nearly $1.5B across nine funds and written early checks into Warby Parker and Casper. The title of the episode is "We Don't Fund Good Companies," which sounds like a provocation but turns out to be one of the cleaner frameworks for thinking about early-stage bets we've heard in a while.
Ben Lerer's argument is this: a "good company" is a coherent, sensible business with a reasonable market and a reasonable team. Those companies get funded, find product-market fit, and plateau. The venture power law doesn't care about reasonable. It cares about outlier outcomes. So Lerer Hippeau deliberately looks for founders who seem a little unhinged about their idea, people who can't be talked out of it, because that stubbornness is the one input you can't manufacture later. The flip side is that "crazy" without conviction is just noise, so the IC process at Lerer Hippeau is built to separate the two.
Conviction over consensus in the investment committee
The most structurally interesting thing Lerer describes is how his IC runs. Most early-stage IC processes try to build consensus: you present, the partners debate, you look for rough agreement before writing the check. Lerer pushes against this.
His framing: consensus filters out the deals where someone in the room has a strong negative reaction. But those deals are often the ones where someone else in the room has a strong positive reaction. If you require consensus to write a check, you're systematically cutting the high-variance bets that actually move the portfolio.
His solution is to run on conviction, not consensus. One partner needs to own the bet. That partner has to be able to articulate, specifically, what they believe that the market doesn't yet believe. If they can do that, the check gets written, even if others in the room are skeptical. The skeptics don't veto; they're on record with their reservations and the deal moves.
This has a consequence most firms don't talk about: it puts the IC member's personal track record on the line more visibly. Lerer acknowledges this. He says he wants to be the "worst investor at his own fund" because that would mean the partners he's brought in are generating better returns than he is, which is the goal of firm building. The incentive isn't to protect your own batting average; it's to build the best collective portfolio.
For B2B founders watching how their investors think, this matters. A firm running conviction-based IC is going to back you harder when the middle-of-the-night doubt sets in, because the partner who wrote the check has publicly staked something on you. A consensus firm might be friendlier in the pitch, but quieter when things get hard.
The Peloton miss and what it reveals about process
Every honest investor has a miss story. Lerer's Peloton miss is worth pausing on because of why it happened.
Lerer's characterization: the deal had the kind of contrarian signal that should have excited them. Connected fitness hardware at premium price points, targeting a consumer who would buy a $2,000 bike for home, was exactly the kind of bet that looks absurd until it doesn't. The market-size skepticism was loud. The "who spends that on a bike?" objection was everywhere. Those are often the conditions under which you find power-law outcomes.
But the process failed. The reservations in the room were too heavy, and the conviction of whoever wanted to write the check couldn't survive the weight of the skepticism. That's a consensus problem. The deal died in committee rather than getting owned by a specific partner.
The lesson Lerer draws isn't "we should have been smarter about connected fitness." It's that the IC structure at the time allowed collective doubt to kill individual conviction. The process reform that followed was structural, not behavioral.
For founders who've had deals die in committee, this framing is useful. It's rarely about the idea being wrong. It's often about the IC architecture at that specific firm. A "no" from a consensus firm tells you almost nothing about whether your idea is good; it tells you whether it generated enough agreement in one room on one day.
AI-native founders vs. domain experts
Lerer addresses the debate that's running through every early-stage firm right now: do you back the AI-native founder who moves at a terrifying pace and has no industry baggage, or the second or third-time operator with deep domain knowledge who might move slower but knows exactly where the bodies are buried?
His answer isn't a silver bullet, which is honest. The case for the AI-native founder is that speed and technical literacy compound fast. A founder who can deploy an agent stack in 48 hours and iterate on it daily has a leverage advantage that domain expertise can't always overcome, especially in markets where the old players are slow.
The case for the domain expert is durability. Knowing why every obvious solution failed before, having the relationships to navigate the sales cycles, understanding regulatory exposure — those things show up later in the company's life when the initial speed advantage has normalized.
Where Lerer lands: he wants the crazy domain expert. Someone who has spent ten years in a market, knows every counterargument cold, and is still convinced there's a fundamental break possible. That combination of stubbornness, expertise, and timing awareness is rare, but it's what he's actively looking for. The AI-native founder who's never touched the industry is a bet on speed alone, which is fragile.
This has a direct read-through for how B2B founders should position themselves. If you're pitching a vertical SaaS play to an early-stage firm, your unfair advantage isn't the AI infrastructure you're building on, it's the ten years you spent watching the legacy tools fail in ways only insiders understand. The technical architecture is table stakes; the judgment about where the breaks are is the asset.
How to win competitive deals as a smaller firm
Lerer Hippeau is not Andreessen. It can't win on check size or brand alone, especially as the funding environment has gotten more competitive at early stages. Lerer is direct about this: smaller firms have to win on relationship velocity and specificity of value, not on signaling.
What that means in practice: getting to a founder before the rest of the market does, building a real relationship before the round opens, and being specific enough about how you can help that the founder has a reason to take your check over the brand-name alternative.
The operator background is part of this. Lerer built Thrillist into a media company before becoming an investor, which means he can talk to media founders, consumer founders, and anyone navigating the editorial-to-commerce transition with more than just pattern matching. He's been in the building. That's the kind of specificity that beats a $500M fund's cold deck when the founder is deciding between term sheets.
The broader point here: differentiation at the firm level and differentiation at the individual founder level work the same way. The firms that win the best deals are visible, specific, and already known by the founders they most want to back. Waiting for the inbound is a losing strategy for a smaller firm competing against established brands.
This is a pattern we see mirrored in how B2B founders build pipeline on LinkedIn. The operators we work with who generate consistent inbound aren't waiting for buyers to find them through search; they're showing up in the conversations their buyers are already having, with a point of view specific enough that the buyer has a reason to care. You can read more on how that plays out at the engagement layer in our piece on why B2B founders should comment on LinkedIn posts.
What "crazy" actually means
The headline deserves more precision than it usually gets. Lerer isn't saying he wants chaotic founders or founders who are bad at execution. The "crazy" he's describing is epistemic stubbornness about a specific market belief.
The founder who has done the work, looked at the evidence, heard every objection, and still believes there is a fundamental break available in the market, that founder is unusual. Most people who hear enough objections update toward the conventional view. The founders who update toward stubbornness are the ones who were seeing something others weren't, or they were delusional. The IC process is supposed to separate those cases, but it never does it cleanly.
What's interesting about this framing for B2B founders: it's essentially a description of how strong content works too. The operators who build real inbound through LinkedIn aren't posting conventional takes. They're posting the thing they believe that most people in their market would push back on. The content that converts is the content that sounds wrong at first pass, then right by the end of the post. It's the written equivalent of what Lerer is looking for in a founder.
That parallel isn't accidental. The founder whose content consistently challenges the safe consensus view is usually the founder who also has the epistemic stubbornness Lerer is describing. It's one signal.
“The deal that dies in committee because of collective doubt, not because the idea was wrong, is the one worth reverse-engineering.”
What funded B2B founders should take from this
Three things stand out from a practical standpoint.
First, how your investors are structured matters more than the dollar amount on the check. A consensus-driven IC is a fragile ally. When your growth stalls, you want a partner who owns the bet publicly and has an incentive to defend it, not a committee that can point fingers in multiple directions.
Second, the AI advantage is real but perishable. Lerer's "AI-native versus domain expert" framing implicitly signals that technical speed alone is not the durable moat VCs are looking for. The founders getting the highest-conviction bets right now are the ones pairing AI execution speed with deep enough market knowledge to know which problems are worth solving.
Third, how you show up before the raise matters as much as the pitch. Lerer Hippeau wins deals by building relationships before rounds open. The implication for founders is symmetric: the investors who will matter most to you already have a view of who you are before you ever send an intro email. What they see on LinkedIn, what they hear from the founders in their portfolio, and what they read from you in public determines whether they take the call. We've tracked how this compounds over time in our analysis of founder-led brands and LinkedIn inbound.
The episode is worth your hour. Find it on YouTube, Apple, or Spotify. Ben's framing of the IC process alone is worth the listen for any founder thinking about their next raise.
Frequently asked
Lerer's argument is that a 'good company' is a sensible, coherent business with a reasonable market. Venture economics don't reward reasonable outcomes. Lerer Hippeau deliberately seeks founders with unusually strong conviction about a market break that most people would push back on, because that stubbornness is the input that separates power-law outcomes from median ones.
