Decision debt, and how to find yours

Somebody has put a slide up with two numbers on it. One comes from the marketing platform, one from the CRM, and they describe the same quarter. The meeting spends eleven minutes on why they differ. Everybody is well prepared, nobody is being difficult, a reasonable explanation is found, and the agenda moves on.

I have sat in that meeting many times, in companies that had nothing else in common. The explanation was found last quarter too, and it will be found again in March, because nothing in the room has the authority to end the question. The two numbers differ because two teams were allowed to define the same word differently, years ago, by nobody in particular.

The eleven minutes are not the cost. The cost is that no figure leaving that room can be defended by anyone who was not in it. Multiply that by every meeting where a question gets explained rather than settled and you have the real shape of the thing. Who is allowed to decide this. What we mean when we say lead. What actually gets rewarded in December. How long it takes us to admit we were wrong. Nobody refuses to answer those. They never arrive in a form where an answer is required, and the cost of not answering lands somewhere else. Usually on marketing.

I call this decision debt: the accumulated cost of the decisions an organisation never made.

Why debt is the right word

Engineers have had a word for their version of this for thirty years, and the word did real work. Technical debt gets written down. It sits in a backlog, it has a number, somebody argues for it at planning. It still compounds, but the organisation can see the balance.

Decision debt has no ticket, because the nature of the thing is that it was never raised. So it arrives disguised: a reporting problem, an alignment problem, a culture problem, a tooling problem. Those are symptoms, and reorganising around a symptom is how the same company rebuilds its marketing function twice in five years and lands back in the same argument.

The metaphor holds further than it looks, because an unmade decision has a principal and an interest, and the two are nothing alike. The principal is the decision itself, and it is almost always cheap: an afternoon, once the right people are in a room and one of them can close it. The interest is what you pay every quarter the question stays open. A campaign rebuilt because the brief was reinterpreted on the way through. A hire delayed two quarters because nobody owns the headcount. An AI pilot producing more content, faster, in a direction nobody validated.

Organisations do not avoid these decisions because they are expensive. They avoid them because they are uncomfortable. Somebody loses scope. Somebody's number gets worse before it gets better. So the company pays the interest instead, which is diffuse, deniable and charged to nobody in particular. Rational for every individual involved. Ruinous for the business.

Four ledgers

Because the debt is unrecorded, the first useful act is to record it. Four ledgers cover most of what I have found across seven industries and three continents, and each has a test you can run this week without asking anyone's permission.

Principal · one conversation, once

Interest · every quarter it stays open

1. I

   Rights

   Who is allowed to decide, and who can overturn it.

   TestAsk five people, separately, who owns a decision currently in flight. More than one name means it is unowned.

   InterestIt gets decided repeatedly and held never. Every forum reopens it, believing it is entitled to.

   SettleWrite one name against the decision, and name who may overturn them. Say it in the forum that keeps reopening it, not in a document nobody reads.

2. II

   Definitions

   What the words mean, identically, in every system and every meeting.

   TestPull the same metric from two systems on the same morning. If the reflex is to explain the gap rather than close it, the definition is unsettled.

   InterestNo number survives a second department, so finance cannot defend a budget line with it.

   SettleAgree one definition in a room that includes sales and finance, write it into the system that reports it, and delete the rival versions. A definition that lives in a slide has not been settled.

3. III

   Incentives

   What the organisation pays people for, which is not always what it asks them for.

   TestOpen the objectives your team will actually be appraised against in December. If the priority you have repeated all year is not written in them, it is not a priority. It is a preference.

   InterestThe strategy gets restated every quarter and nothing moves. People are not resisting it. They are correctly following the version of it that pays them.

   SettleChange what December measures, not what January announces. One objective rewritten is worth more than a year of restating the priority.

4. IV

   Reversibility

   How long it takes to unmake a decision once the evidence says it was wrong.

   TestFind the last decision that was reversed on evidence and count the months between the evidence arriving and the decision changing. If nobody can name one, the answer is not zero. It is that reversals do not happen here.

   InterestThe wrong decision keeps costing full price for every one of those months. And because everybody has watched how long it takes, they stop bringing evidence at all, which makes the next one slower still.

   SettleSay, at the moment you take the decision, what evidence would make you reverse it. A decision with no stated exit gets defended instead of reviewed.

The first three are about settling things. The fourth is about unsettling them, and it is the one people skip. Without it an organisation does not stop taking on decision debt. It simply takes it on with more confidence, and holds the wrong answer as firmly as it would have held the right one.

Where to start, which is not a sequence

The four interlock in a way that defeats any clean order. You cannot assign ownership of something nobody has defined. You cannot define it without somebody empowered to end the argument. Neither survives if December rewards the opposite.

What works is smaller than a sequence. Take one decision that matters, borrow just enough authority to close it, and use the fact that it is closed to argue for the next one. A single agreed definition of a qualified lead, signed by sales as well as by marketing, is worth more as a precedent than as a definition. It proves this organisation can settle something, which is not a thing everyone currently believes.

That is why the work takes six months rather than six weeks, and why the dashboard is never the hard part. The dashboard is a weekend. The six months are the conversation that makes the dashboard mean anything.

What paying it down looks like

It presents badly. There is no launch, and the deliverable is an argument that stops happening. For a while it looks as though nothing is being produced, because what is being produced is the absence of future rework, and nobody has ever built a slide for that.

Then the downstream gets suspiciously cheap. Briefs stop being reinterpreted. Reviews start at the discussion instead of at the definitions. A model that was producing noise starts producing decisions, because for the first time it is pointed at something the company agrees on.

Marketing pays a disproportionate share of this interest for a structural reason. It sits furthest downstream of the decisions and it is the first function asked to show a number. Every unsettled question upstream arrives eventually as a campaign carrying weight it was never designed for. That is why so much marketing looks like a marketing problem and is not one.

Find the decision nobody would make. Make it. Then hold it long enough that the organisation forgets it was ever in question.