Most boards are not hostile to new bets. They are, however, structurally designed to strangle them.
The typical board governance model was built to steward a going concern — quarterly reporting, roadmap oversight, milestone tracking. Applied to a portfolio of new bets, those same instruments produce exactly the wrong incentives: they reward legibility over learning, and they punish the honest report that a bet is not working before the sunk cost is large enough to justify killing it.
What a new bet actually needs from governance
**Staged capital, not annual budgets.** Each bet gets a small first tranche, released against a hypothesis. Additional capital is unlocked only when the previous tranche's evidence clears a pre-agreed bar. The board's job is to approve the stage, not to grade the roadmap.
**A shutdown criterion written before the money is released.** The single most useful sentence in new-bet governance is: "We will stop this if, by [date], we do not see [evidence]." Written before the work starts, when reason is cheapest. Held to, afterwards.
**A learning-rate report, not a milestone report.** Every quarter, the portfolio reports what it knows now that it did not know last quarter, and what decisions changed as a result. If nothing changed, the portfolio is not learning; it is performing.
Boards do not need to become product experts. They need to become good customers of a well-instrumented learning process.
Where most portfolios fail
Not at the beginning. At the moment the first bet needs to be killed. The organisation has not rehearsed shutdown. The team feels the ending as a personal failure. The board reads it as an execution failure. The instrument that was supposed to protect the portfolio — the shutdown criterion — is quietly renegotiated, and the bet lives on, consuming attention no one can afford, until the next reorganisation quietly buries it.
The fix is to shut the first bet down in public, on the criterion, before the sunk cost demands otherwise. Once. Everything after that becomes cheaper.