VCs as Herd Animals

The incentive that produces the herd

Ryan Petersen explains venture herd behavior as a property of the job rather than of the people who hold it. He lists the attributes of being a venture capitalist: it pays very well, a senior partner has no boss, there is no fixed schedule, fund fees are locked for roughly ten years, and it is very hard to measure whether an investor is any good on any reasonable timeframe.¹ Given that, he asks how a VC actually gets fired, and finds two ways: a scandal, or letting everyone at the firm conclude they are dumb or doing bad deals. The consequence he draws is that a deal can be good and still get an investor fired if the partners merely think it is bad, so perception on a short timeframe, not the eventual outcome, is the operative risk.

From that premise Petersen describes the behavior he says follows. Investors seek consensus, channel-check, and spot-check a deal before risking looking foolish in front of their own partners. His sharpest claim is that most VCs collude more with competitors than with their own partners, because they need to verify a deal is good with rival firms before bringing it internally.¹ He locates the mechanism most visibly one level down, among associates who he says run cross-firm chat groups and trade weekly roundups of the founders they have met, reasoning that they need to give to get, and that an associate is in more direct competition with the other associates at their own firm than with rivals.

The demonstration Petersen offers

Petersen supplies a firsthand example rather than an argument. He says he told one fund mid-process that he was going a different direction, though he had not actually chosen anyone, and within an hour three other funds he was still talking to called to say they had heard he had picked a lead.¹ He treats the speed of propagation as evidence that a single read travels instantly across the network. He names two exceptions he sees to the pattern. He credits Keith Rabois with sense-checking deals against friends and treating their agreement as a sign he is not doing his job, the deliberate inverse of consensus. And he argues that richer investors make better investors, citing Sequoia's focus on upside maximization, asking "how big could this be?" rather than downside protection, because a firm not afraid of losing its next fund is free to chase outliers rather than herd toward safety.

What Petersen says founders should do

The principle is framed by Petersen as a warning to founders who are raising. Because a lukewarm read propagates through the associate network almost immediately, he argues a founder should never test the market with only one or two meetings, since a weak early signal can poison the rest of the process.¹ His stated goal for a founder is to find the single outlier investor who values the company more than everyone else, which he says consensus dynamics work against unless every node in the network is already enthusiastic. This connects to his advice in Never Share Your Metrics, where the same concern about information leaking through the herd shapes what a founder discloses during a raise.

Petersen also uses the principle to explain returns. He argues the herd is the structural reason the Venture Power Law is left on the table, because the consensus deal is by definition not the contrarian outlier that returns a fund, an idea he links to the kind of contrarian question that consensus suppresses.

Where the framing is contested

Two limits attach to Petersen's own telling. He is now a partner at Founders Fund, the firm he holds up as an exception to the herd, so the account is partly self-flattering, and whether that firm's contrarianism is structural or driven by particular personalities is not settled by his argument. And the word collude is doing rhetorical work: channel-checking and reference-trading are also ordinary diligence, and the strong version, that investors side with competitors over partners, is asserted from his own experience rather than measured.