Start With What Can't Be Reversed

The recovery engagements I walk into tend to start the same way. Three or four things are broken at once, and someone wants to know what gets fixed first.

The instinct is to go by volume. The squeakiest problem usually gets attention first. It feels responsible. Most of the time, it's the wrong call.

How the quiet work loses

Years ago, I worked in a delivery environment where a client leader micromanaged the team closely enough that whatever they wanted that week became the priority automatically, without anyone deciding it that way on purpose. It felt responsive. Every sprint, the loud, visible ask won. The quiet one, paying down shortcuts the team had already taken to hit earlier dates, kept losing. The team didn't set out to bury that work. It just never beat whatever felt more urgent that week. By the time the debt itself got treated as the emergency, fixing it meant tearing out things that had already shipped, in front of users who now depended on them.

That's the pattern I keep finding in recovery work. The fire that gets fixed first is usually the loudest one, not the one closing a door permanently.

The half of the quote everyone skips

Jeff Bezos gave this distinction a name in Amazon's 2015 shareholder letter, in a passage that often gets half-quoted. He split decisions into two kinds. Type 1 decisions are, in his words, "consequential and irreversible or nearly irreversible, one-way doors," and they deserve slow, careful process. Type 2 decisions are reversible, and they don't. His complaint wasn't that companies take too many risks. It was that they apply Type 1 caution to Type 2 decisions, and the result, in his words, is "slowness, unthoughtful risk aversion, failure to experiment sufficiently." The irreversibility warning is the half that gets quoted. The half that gets skipped is the one that matters more once you're in a recovery.

In a healthy company, the failure mode Bezos describes is paralysis: everything gets treated like a one-way door, so nothing moves fast. In a recovery, the failure mode inverts. There's no time to sort decisions properly, so everything gets treated like a two-way door by default, including the ones that aren't. The genuinely irreversible call, the one that isn't being shouted about, gets the same five minutes of attention as everything else on the list.

Why it compounds

There's a reason this compounds instead of averaging out. Barry Boehm's cost-of-change research from 1981 found that the cost of fixing a decision rises by roughly an order of magnitude with each phase it survives past its origin. The exact multiplier has been debated since; it moves with project type and shrinks under short-cycle delivery. But the direction has held up: a decision left in place gets more expensive to reverse, not less, the longer it sits unexamined. Solution debt is a decision, not an accident. It gets a little more permanent every sprint it goes unnamed.

The scale here is bigger than a handful of unlucky teams. The Standish Group's last full CHAOS study put roughly half of IT projects in the "challenged" category, over budget, late, or short of scope, with less than a third landing as outright successes. A lot of delivery work eventually turns into recovery work, which is exactly why the triage question deserves more rigor than "who's loudest."

The habit

The fix is a habit, practiced before the war room starts: before reacting to whichever fire is loudest, ask what closes a door if it's left alone for another week. A slipped date is a two-way door. It reopens the moment you have more capacity. A client relationship that's gone quiet, a vendor swapped under pressure, a piece of debt that isn't tracked on any roadmap: these often turn out to be one-way doors mistaken for two-way ones, right up until they aren't.

What can be reversed can wait a day.

Sorting out which is which, before the fixing starts, is most of the job.