Say a chief product officer chairs a committee that decides which new ideas get money. Five senior leaders sit on it. The rule is one everybody can recite from memory: nothing is funded unless a majority is behind it. Three years in, the committee has approved forty proposals. Each had a business case that held together, a sponsor who presented well, and nobody in the room who objected hard. Ask the chair to name a proposal the committee turned down that later became somebody else's product line, and you would probably get a long pause, mostly because nobody has ever gone back to look.

That voting rule feels like governance, and for a good share of what crosses an executive desk it is exactly that. A paper on venture capital points somewhere else. The way I read it, a voting rule is a decision about which mistakes you are willing to make, and the authors' model says the right answer changes with the kind of bet on the table.

What venture firms do at the front door

The paper, by Andrey Malenko, Ramana Nanda, Matthew Rhodes-Kropf, and Savitar Sundaresan, opens with a survey. The authors asked managing partners at 55 of the largest U.S. venture firms how their investment committees vote, and 35 answered.1 Sixty percent of them said that for seed stage deals they use what the authors call a champions rule, under which, in the paper's words, “a single partner can go ahead and do the deal regardless of what the others feel.”1 Another 30 percent use a version in which one champion is enough unless someone vetoes.1 For early stage deals, usually a startup's first major round, the share using the champions rule falls to 20 percent, and the authors report that the same firms use a more conventional majority rule for later stage investments.1 In a follow up survey that 19 of the firms answered, the reason they backed most strongly was to “catch outliers.”1 The paper has been accepted at The Journal of Finance, according to the Imperial College London repository, though the version I am describing is the 2023 working paper.2

What a majority screens out

The authors build a model to explain the practice. Each partner sees one slice of a project, and an exceptional project can be strong on some slices and flawed on others. In the paper's words, rules that require many votes “result in the undertaking of projects that tend to be good on many dimensions but are likely to miss those that are exceptional on few dimensions but are mediocre elsewhere.”1 The paper quotes Marc Andreessen on what happens when partners' scores are combined: "you get the mush in the middle, with no big flaws but no great strengths."1

A committee that requires broad agreement will fund what survives broad scrutiny. If you want the exceptional bet, someone has to be allowed to be wrong alone.

The authors then run a numerical example with five partners, project qualities drawn from a power law, and an even chance that any project fails regardless of its quality. In that example, a fund of 25 independent investments has a 52 percent chance that at least one returns ten times the money or more under the champions rule, against 16.1 percent under majority rule.1 That is a modeled result, not observed returns, and the authors note that one of their simplifying assumptions exaggerates the gap between the rules.1

The condition that decides which rule wins

The champions rule does not win everywhere, and the authors say so. In their model it dominates other voting rules only where the underlying characteristics have heavy right tails, meaning a few outcomes dwarf the rest.1 Where tails are light, a superstar project is typically driven by many characteristics that are each good, and more consensus is the better rule.1

My guess is that most corporate funding committees run a single rule for everything, usually a majority or something close to unanimity, because it looks like diligence. For a plant expansion or a system migration, where the downside is a large and visible loss and the upside has a ceiling, that rule earns its keep, since five readers catching one fatal flaw is the whole point. For a portfolio of small experiments where one success may be worth more than the rest combined, the same rule is likely to screen out what the committee was formed to find. A proposal that is brilliant on one dimension and awkward on two others collects three skeptical votes, and the committee has followed its own rules perfectly.

The authors make a version of this jump themselves, saying the logic could apply to hiring and research spending wherever outcomes have a long right tail.1 They also give a reason companies rarely do it: when committee members have a personal stake in a project, a lone champion can push through bad ideas, and that cost can outweigh the benefit of catching an outlier.1

My read of the paper is that it argues for matching the rule to how the payoffs are spread. I would add that the evidence comes from venture capital, where heavy tails are the premise of the business. An operating company's innovation budget may not have them, and finding out is the first job.

A way to find out which rule you are running

Pull the last two years of approvals. If every item had a defensible case and none surprised anyone, the committee is selecting for the absence of objection. Then ask whether the money in that docket tends to produce a few large outcomes or many modest ones. If it is the first kind, I would try a small budget that one named executive can commit without a vote, and keep the committee for the large commitments that are hard to reverse. I wrote about a neighboring question in how delegation should vary with the type of decision, and this is the same instinct applied to the vote itself.

Think of the last proposal your committee turned down by a split vote. Does anyone know how it turned out?