Decision Elimination: The Largest Hidden Lever on Transaction Cost

🧠 Free Recall ❓ Quiz 🔍 Cues 🔮 Simulation 🥷 ShadowBox 📐 Encode Cards

A decision doesn't feel like a transaction. It feels like thinking — the valuable part, the thing we agreed earlier to protect rather than attack. But in flow terms a decision is a unit of work like any other: it passes through a server, waits in a queue, and levies overhead on both entry and exit. Once you see it that way, the case for eliminating decisions — not merely making them faster — becomes hard to argue with.
Start with the server. Every decision a knowledge worker makes routes through the same scarce piece of hardware: the prefrontal cortex's executive function, a low-throughput, metabolically expensive resource that operates more or less serially. Cognitive load theory (Sweller) puts a hard ceiling on it — working memory holds only a handful of elements at once. So the thing doing the deciding is, in systems-thinking terms, the constraint. And the first rule of a constraint is that any minute you can avoid spending on it is worth far more than a minute saved anywhere else.

Now the queue. Reinertsen's central, counterintuitive lesson in The Principles of Product Development Flow is that queues — not capacity — are the dominant cost in development work, and they stay invisible because organizations track how busy people are instead of how much work is waiting. The relationship between utilization and queue size is nonlinear: as a server approaches 100% utilization, wait time doesn't rise gently, it bends toward infinity. The human decider runs permanently near full utilization — we are essentially never not deciding. We therefore live on the steep part of that curve, where each additional decision inflicts disproportionate delay on every other decision stacked behind it. Pending decisions are work-in-process held in your head, accruing a cost of delay the whole time they sit there unresolved.

Reinertsen's batch math sharpens the point. Economic batch size is set by the balance between transaction cost (the fixed overhead of processing a batch) and holding cost (the cost of delay while work waits). Drive the transaction cost down and the economic batch shrinks, which speeds flow everywhere. A decision sits on both sides of that equation at once: it is a transaction cost levied on the work it gates, and it carries its own transaction cost — the ramp-up to load context, the residue on the way out, the fatigue it leaves behind. Lowering the count of decisions is the most direct way to lower the aggregate transaction cost of a working day.

Put it together and three mechanisms compound. Eliminating a decision removes its transaction cost outright — you cannot beat zero, and no amount of optimization gets you there. It also unloads the constraint, lowering effective utilization, which — because of the nonlinear queueing curve — buys outsized relief for every other decision still waiting. And a less-loaded decider is a sharper one, so the transaction cost of the decisions you do keep falls too, shrinking economic batch size and quickening flow downstream. One move, three returns.

Elimination also has a richer toolkit than it first appears, and most of it maps onto a single idea we already surfaced: pay a fixed cost once to kill a recurring one. Decide once — convert a repeating decision into a standing policy or heuristic, so the marginal cost of each future instance drops to near zero (this is exactly the "record the decision and rationale" card, doing economic work). Delegate — move the decision to a different server. Constrain — shrink the option set before you ever look at it. Default — make the common path the no-decision path, so the choice only surfaces when something is genuinely off.

This is why, of all the cards on the table, the decisions you stop making are among the highest-leverage of all. Speeding a decision saves its transaction cost once. Eliminating it saves that cost forever, frees the constraint, and — because you're sitting on the steep part of the queueing curve — purchases nonlinear relief for everything still waiting to be decided. The fastest decision is the one you've arranged never to have to make.

Unstarred

Dunk (exclude from study)

Decision Elimination: The Largest Hidden Lever on Transaction Cost

Review Questions

Todos

Sensemaker