Decisions that compound
In the last essay I argued that most institutions do not compound — they re-discover. The question asked in March gets answered in March and forgotten by May, and when its cousin arrives in September, the work starts from zero. Re-discovery is a sawtooth. Learning is a staircase.
This essay is about what the staircase is made of.
An idea every credit union already understands
There is no industry on earth better positioned to understand what I am about to say, because credit unions teach it to members every day: the difference between simple and compound interest. Simple interest pays you on the principal, over and over. Compound interest pays you on everything you have already earned. Over one year the difference is small. Over twenty, it is the whole story.
Decisions work the same way — and almost every institution is earning simple interest on them.
A decision earns simple interest when it produces its outcome and nothing else. The pricing committee decides; the rate changes; the moment passes. The work that went into it — the analysis, the debate, the reasoning, the alternatives considered and rejected — evaporates. Next year's pricing committee starts from the principal again.
A decision compounds when it leaves something behind that makes the next decision better: its reasoning, its evidence, its outcome, all findable when the next question arrives. The 2024 pricing decision becomes part of the foundation the 2026 one is built on. The institution is now earning on everything it has already recorded.
What a decision must leave behind
For a decision to compound, three things about it have to survive the meeting where it was made:
Not just what was decided, but why — what question it answered, what the alternatives were, what tipped the balance. Most institutions preserve the what in board minutes and the why in someone's memory.
The numbers that supported it, and where they came from. When next year's committee asks “what were we looking at when we decided this?”, the honest answer in most institutions is a hunt through email attachments.
What actually happened. This is the rarest of the three, because it requires someone to close the loop months later — connecting a result back to the decision that produced it. Without the outcome, an institution cannot even tell which of its past decisions were good.
Where all three survive, something happens: decisions start referencing each other. The deposit-pricing decision cites the liquidity analysis; the branch decision cites the member-migration finding; the 2026 review cites the 2024 rationale and can say, with evidence, it worked or it didn't, and here is why. That web of connected reasoning is what institutional judgment actually is — and today it lives almost entirely in the heads of long-tenured people.
Decision debt
There is a name for the opposite condition, borrowed from software: debt. Every significant decision made without leaving its reasoning behind is a small loan taken out against the future — repaid, with interest, the next time someone has to reconstruct why things are the way they are. Why is this product priced this way? Why did we exit that community? Why does this policy exist? In an institution carrying heavy decision debt, the answer to all of these is the same: we'd have to ask around.
The cruelest part is who pays. Decision debt comes due precisely when the people who made the decision have moved on — which, over the multi-decade life of a credit union, is a certainty, not a risk.
Compounding needs every level
One more condition, and it is the one hierarchies find hardest. Interest compounds in a single account; decisions compound across an institution — which means they have to travel between levels. The board's strategic choice has to reach the front line as a changed action; the front line's ground truth has to reach the board as evidence. When each level decides from its own picture, the levels quietly work against each other, and the compounding breaks exactly where the org chart draws its lines.
An institution that decides together moves together. An institution that decides in layers pays simple interest at every layer.
The test
Here is the compound-interest version of the three questions from the last essay. Pick one significant decision your institution made two years ago, and ask: could a capable person, starting today, find its reasoning, its evidence, and its outcome in under an hour — without interviewing anyone?
If yes, that decision is compounding. If no, it was an event. It happened, and then it was gone, and everything it taught you is earning nothing.
Member relationships are measured in decades, and so is the return on an institution that lets its decisions build on each other. The first institutions to move from simple to compound will not look dramatically different in year one. By year ten, the gap will not be closable.
Which raises the last question of this series. If an institution's knowledge is going to accumulate — decision upon decision, answer upon answer — it will need somewhere to accumulate in. Some structure. Which is another way of asking the strangest and, I think, most important question of the three:
Does your institutional knowledge have a shape?