Every decision was defensible. The totality was a failure.
Years ago, at a prior institution, I watched a capable senior leader make a series of decisions that were each, taken alone, reasonable. Sound analysis. Right stakeholders in the room. Defensible on the day. And when those decisions merged in the real world, they produced an outcome nobody had chosen and everybody had built.
The hard part came after. I had to walk back through the chain and explain how each decision had quietly narrowed the options for the next one — how a sourcing call constrained an architecture call, how the architecture call locked in an operating assumption, how the assumption aged badly and took three downstream commitments with it. No single decision was wrong. The system of decisions was.
That experience has stayed with me for a simple reason: nothing in how we run large organizations would have caught it. We review decisions one at a time. We reward decisiveness. We document outcomes. We almost never examine what a decision does to the decision space around it. This note is my attempt to name that gap.
Decisions compound.
Every consequential decision changes the conditions under which future decisions get made. It forecloses options, creates dependencies, sets precedents, and allocates attention. That's not a side effect — it's the primary long-run consequence, and it usually dwarfs the direct one.
Which means organizations accumulate decision debt the way codebases accumulate technical debt. Shortcuts in how decisions get made — unclear ownership, unstated assumptions, horizons that stop at the current quarter — don't fail immediately. They compound quietly, and the interest comes due later, paid by people who weren't in the room.
Here's the distinction the whole argument turns on: decision quality is not decision integrity. Quality asks whether this call was sound — right information, right analysis, right judgment. Integrity asks whether this call holds together with everything upstream and downstream of it: whether it honors the constraints it inherited and whether it's honest about the constraints it creates. An organization can be full of high-quality decisions and still be drowning in decision debt. That was exactly the situation I watched unfold.
Decision Risk deserves its own layer.
We've professionalized nearly every risk that matters to a large institution — credit, market, operational, technology, model, third-party. Each has owners, appetite statements, and controls. But the risk that a series of individually sound decisions compounds into an unsound position has no name, no owner, and no framework. It shows up in the post-mortem, never in the register.
I've started calling it Decision Risk, and I think it's the most under-managed risk in modern organizations — and getting worse. AI is collapsing the cost of analysis, which means decision velocity is going up everywhere. More decisions, made faster, by more actors, with fewer natural pauses. Compounding accelerates with volume. The organizations that thrive won't just make better individual calls; they'll manage the integrity of the whole decision system.
What a decision should pass before it ships.
The practical core of Decision Integrity is five tests. None require a committee. All require honesty.
Taking on decision debt isn't the sin. Hiding it is.
None of this argues for slow decisions or decision-by-committee. Speed matters, and sometimes the right call is to knowingly take on decision debt — lock in a vendor before the strategy is settled, ship the interim architecture, accept the precedent. Engineering teams do this with technical debt all the time, and the good ones do it well because they do it in the open: the shortcut is named, logged, and scheduled for repayment.
Decision debt deserves the same treatment. The failure mode isn't the shortcut. It's the unrecorded shortcut — the constraint nobody wrote down, discovered two years later by a team that can't understand why their options are so narrow. Deliberate debt is a tool. Silent debt is a trap.
Leaders are judged on the decision systems they leave behind.
We evaluate leaders on results, and results matter. But results are partly luck and often lag. The more durable measure is what a leader does to the organization's capacity to decide well after they're gone: whether ownership is clear, whether assumptions get written down, whether debt is deliberate, whether the decisions they made left the option space wider or narrower for the people who follow.
That's the discipline I'm calling Decision Integrity. This note plants the flag; the full framework is in the works — how to score it, how to review for it, and what a decision-mature organization looks like in practice. If this names something you've lived, I'd like to hear the story.