Decision Architecture

Capital decisions don't fail because of bad judgment. They fail because the architecture isn't there.

Most $25M–$150M companies don't have a decision-making problem. They have a decision architecture problem. They have processes, controls, and approvals — but no architecture. Capital decisions get made in Slack DMs and the back seat of an Uber. Decision rights live in someone's head. Capital approval logic gets reconstructed after the fact, when the wire has already gone out.

Decision Architecture is three components: Decision Rights, Decision Logic, Decision Review. Each defined and defensible. Each a discipline, not a tool. Each forces a question that most companies cannot answer in writing.

Decision Rights

Who has authority to commit capital, at what threshold, with what review?

Decision Rights is the written matrix that names who can commit capital, at what dollar threshold, with what level of review attached. It is not an org chart. It is not a delegation of authority memo from 2019. It is the operating document that governs how capital actually moves through the business — and who is accountable when it moves badly.

Without it: a $400K vendor contract gets signed by an SVP who believed she had authority because nobody told her she didn't. Three months later, the CEO discovers the spend in a board deck and asks who approved it. Nobody can produce the answer in writing. The decision was made; the architecture wasn't.

Could you produce your decision rights matrix above $250K — in writing — in the next 24 hours?

Decision Logic

What assumptions drive the choice? What's the downside? What are we walking away from by saying yes?

Decision Logic is the documented thinking behind a capital commitment. The base case, the downside case, the assumption that breaks the model, and the explicit articulation of what the company is choosing not to do by saying yes. Most mid-market companies skip the last two. Both matter more than the first.

Without it: a $2.4M acquisition closes on a base case that assumed customer retention at 92%. Eighteen months later, retention is at 71%. Nobody can find the document where 92% was justified. The decision logic was a slide; the architecture wasn't.

On your last three capital decisions over $1M, can you name the single assumption that, if wrong, breaks the thesis?

Decision Review

When do we revisit? Quarterly? Trigger-based? What makes a decision accountable to a future check-in?

Decision Review is the discipline of returning to past decisions on a defined cadence — quarterly, trigger-based, or both — and grading them against their original assumptions. Not to assign blame. To compound judgment. Companies that review decisions get better at making them. Companies that don't, repeat them.

Without it: the same $50M acquisition thesis gets pitched to the board every 18 months. The thesis was wrong the first time. Nobody revisited it. So the second pitch lands on a board that doesn't remember why it killed the first one.

When did your board last revisit a decision made 18 months ago to grade whether the original assumptions held?

Seven questions. In writing.

  1. 1.Can you point to your decision rights matrix in writing — or does it live in someone's head?
  2. 2.When was the last time a capital decision over $250K was reviewed against its original assumptions?
  3. 3.If your CEO left tomorrow, would the decision logic on the last three major capital deployments be reproducible?
  4. 4.How often does your finance team reconstruct the rationale for a capital decision after the wire has already gone out?
  5. 5.Is there a written threshold above which capital decisions trigger a structured review — or is the threshold whatever the room agrees to that day?
  6. 6.Can you name the three operational drivers your forecast actually depends on — or is it built around the chart of accounts?
  7. 7.When did your board last revisit a decision that was made 18 months ago to see whether the original assumptions held?