Reversibility: the dimension most decisions skip

A decision that can be undone in an afternoon should be made in an afternoon. A decision that will take a year and a migration project to undo deserves evidence, alternatives, and a written reason. Treating both with the same process is how organisations end up either paralysed on small things or reckless on large ones.

What makes a decision hard to reverse

• Cost of reversal: the money and people needed to undo it.
• Time to reversal: weeks, quarters, years.
• Migration difficulty: is there a path, or a rewrite?
• Vendor dependency: how much only works with this supplier?
• Data portability: can we take our data, with its meaning intact?
• Contractual constraints: minimum terms, notice, exclusivity.
• Operational dependency: what would have to keep running through the change?

Why we resist a single score

It is tempting to roll these into a number. We don't, unless the underlying information supports it, because the number hides the thing you need to act on. “Reversibility: 6/10” tells you nothing. “Data export is straightforward; the contract has a 24-month minimum; the team has no skills in the alternative” tells you exactly where the risk is and what would reduce it.

Using it in practice

Classify each significant decision as low, medium or high reversibility cost. Match the deliberation to the class. For high-cost decisions, look for a way to make them cheaper to reverse: an abstraction at the boundary, a data export tested before signing, a shorter term at a higher price. Often the best architecture decision is not the choice itself but the work that keeps the choice open.

Some doors swing both ways. Some lock behind you. Know which one you're walking through.