The Mechanism: Cadence Spends Slack
Every decision an organization makes draws on a shared reservoir: the unallocated attention of its senior operators, the uncommitted capital on its balance sheet, and the reversibility of the commitments it has already entered. Raising the operating cadence — shipping more decisions per quarter — increases throughput, but each decision withdraws from that reservoir. The error most boards make is to treat cadence as free, as though running the same machine faster costs nothing but effort. It does not. A faster cadence converts slack into committed positions, and committed positions are precisely what an option is the absence of. The trade-off is therefore not throughput against rest; it is throughput against the future ability to choose.
Why Volume and Quality Diverge
Decision quality is not the average merit of the decisions taken. It is the expected value of the portfolio of bets an organization holds, including the bets it has deliberately not yet placed. An undecided question with real downstream optionality has positive value: it can be resolved later, with more information, against a clearer market. Closing it early to lift the cadence captures the headline win but extinguishes the option premium. This is why a team can post a rising decision count while its actual position deteriorates — it is liquidating optionality to manufacture the appearance of momentum. The two curves diverge because volume rewards closure and quality rewards keeping the right things open.
Model the operator's reservoir as a buffer that refills at a fixed rate — the pace at which information arrives and capacity recovers. Each decision consumes a draw plus a recovery interval before its consequences are legible enough to inform the next move. Push the cadence past the refill rate and the buffer runs dry: decisions begin landing on top of consequences that have not yet resolved, so each new choice is made blind to the last. Quality peaks not where the buffer is full and idle, nor where it is empty and starved, but at the cadence where draw rate matches refill rate and the queue of unresolved consequences stays bounded.
The Failure Mode: Cadence Lock
The characteristic failure is not a single bad call but a regime I would name cadence lock. Once an organization indexes its self-image to decision velocity — once "we move fast" becomes identity rather than instrument — the cadence ratchets and will not come down voluntarily, because slowing looks like decline. The watch-stander's signals appear in a recognizable order:
- Reversible and irreversible decisions are run at the same speed, because the cadence is set centrally rather than by the cost of being wrong.
- Rework rises: choices are reopened weeks later because they were closed before their consequences were observable.
- The slack buffer is structurally empty — there is no senior attention uncommitted to absorb a surprise — so the next shock forces an unplanned, expensive stop.
Cadence lock is dangerous precisely because the leading metric looks healthy. Throughput stays high until the moment the reservoir is exhausted, and then quality fails discontinuously rather than gracefully.
The Decision Implication
The orchestration layer's job is not to maximize the number of decisions but to govern the rate at which the organization spends its slack — to run cadence as a managed variable, not a cultural constant. Two disciplines follow. First, separate the clock from the cost of error: route reversible, low-consequence decisions to a fast lane and let them run at high cadence, while throttling irreversible commitments to the rate at which genuine information arrives. Second, hold a deliberate slack reserve — uncommitted attention and capital that the cadence is forbidden to consume — and treat any quarter in which it falls to zero as a defect, not a triumph. The organization that wins is not the one that decides fastest. It is the one that has matched its cadence to its refill rate, and so still has a move left when the board it cannot see finally turns over.