Decision Debt and What Delayed Choices Cost
Every postponed decision builds up a cost nobody tracks. The clearest case in a small firm today is the AI decision: while the owner waits, the team picks its own tools.
Most leaders track financial debt, technical debt and operational debt, and very few track decision debt. Decision debt builds up every time a choice gets postponed. At first nothing breaks, which is exactly why nobody notices it. While everything looks unchanged, the room around the decision keeps getting smaller.
What happens while a decision waits
Three things happen while a decision sits open:
- Time removes alternatives that were open at the start. A supplier fills its slot, a price offer expires or a good candidate takes another job.
- Stakeholders add expectations about what the answer will be. The longer the team waits, the more people assume the answer they'd prefer.
- Dependencies lock in around whatever the team is doing in the meantime. Work gets built on top of the temporary arrangement until changing it costs real money.
By the end, a decision the firm could once make freely has become one the circumstances force on it. The firm still makes a choice on paper, but most of the options are already gone.
Delay on its own isn't the danger, and some decisions should wait for better information. The danger is delay without anyone noticing that its cost is growing, because nobody put the open decision anywhere they'd see it again.
The AI decision many small firms are postponing
The clearest current example is the decision about AI at work. In many small firms, the owner knows the team should agree on which AI tools are allowed and what can be pasted into them. It feels like a decision that can wait until things are calmer, so it waits.
The team doesn't wait. Take a hypothetical 20-person firm where the owner postpones the AI decision for six months. During those months, a few people sign up for free AI tools on their own accounts. One uses a tool to draft client emails, another uses one to tidy up a price list, and a third pastes in a customer contract to get a summary. Nobody broke a rule, because there were no rules.
All three effects show up in that firm. Time removed the option of starting clean, because the tools are already in daily use. Stakeholders added expectations, because the team now expects to keep the tools it chose. Dependencies locked in, because client work and files sit inside personal accounts the firm doesn't control.
When the owner finally decides, the question is no longer "which tools should we use?". It has become "how do we get our data back out of six tools without stopping the work?", which is a harder and costlier question to answer.
The interest on that delay gets paid in two ways. One is the work of undoing what grew up in the gap. The other is risk, because for six months client information went into tools nobody at the firm had checked. I wrote separately about what your team should never paste into AI, and that list covers the basic rules this firm went without.
Sort decisions by whether they can be undone
Strong operators don't decide everything fast. They sort decisions into two kinds and treat each kind differently:
- Reversible decisions, which they make fast. If the choice proves wrong, the firm can change it next month at small cost, so waiting buys very little.
- Irreversible decisions, which they make carefully. These deserve time, but the time is planned, with a date by which the decision gets made.
Weak systems handle both kinds the same way, and that's where the debt piles up. Small reversible choices wait in the same queue as large ones, and the large ones get no deadline at all.
The AI decision in the example is a useful test, because most of it is reversible. A firm can pick two approved tools and a short rule about client data this week, then adjust both after a month of use. Picking a tool is cheap to undo, while six months of client data spread across personal accounts is much harder to pull back.
A delayed decision works like a loan, and the firm pays interest on it every week it stays open. The cheapest time to decide is usually earlier than it feels.
If your leadership team wants to settle its first AI rules together, ask about the workshop here.