A month after a difficult decision, someone asks why you made it.
You remember the conversation. You remember feeling fairly confident. You can explain why the choice made sense.
But can you separate what you knew then from what you have learnt since?
Consider a hypothetical team choosing between a limited trial and a full rollout of a new customer process. At the time, the trial looks sensible because demand is uncertain. Six weeks later, demand is strong. Suddenly, the full rollout can feel like the obvious choice everyone should have seen.
A short decision record gives you something to return to: what was known, what was assumed and why a particular option was chosen before the result arrived.
Here's how to make that record useful without turning every choice into a paperwork exercise.
Why memory needs something to check against
In a 1975 study, Baruch Fischhoff and Ruth Beyth asked participants to estimate the likelihood of possible outcomes from President Nixon's overseas visits, then later recall their earlier estimates. Their recollections shifted towards what they believed had happened. The researchers found little evidence that stating predictions beforehand protected later recollections from this bias. Read the original study.
That was a study of students and political events, not a trial of leadership journals. It doesn't establish that writing a note improves business results.
It does give us a reason to preserve the original thinking and actually look at it later, rather than rely entirely on a confident recollection.
The practical aim is modest: leave a usable reference. A record cannot prove that a decision was sound, remove hindsight or excuse a warning that was ignored.
Choose a decision worth recording
Start with a choice that involves a meaningful trade-off, uncertainty or commitment.
A limited trial versus a full rollout could qualify. So could changing a customer promise, choosing between competing priorities or proceeding while an important question remains unanswered.
Routine choices don't all need their own document. The record should be proportionate to the consequences and complexity. Keep using any formal approval, safety or specialist-review processes your organisation requires.
There are established examples of this approach in other settings. The UK's Health and Safety Executive asks investigators to keep timely records of significant decisions and their reasons, including decisions not to act. Its guidance stresses explaining the circumstances and information available at the time. This is investigation guidance, not a requirement for everyday management decisions. Read the HSE guidance.
For your team, begin with one consequential choice. Use the place where the work already lives, with access appropriate to the information involved.
Write what you know separately from what you expect
The most useful part of the record may be the distinction between evidence and assumption.
“We have enough demand” sounds like a fact. It might mean customers have made firm commitments. It might mean several people sounded interested in a conversation. Those are different starting points.
Ask:
- What have we actually observed?
- Where can someone check it?
- What are we expecting, rather than knowing?
- Which missing fact could change this choice?
Keep source dates where they matter. A capacity report from last month may not describe the team available next week.
The UK government's AQuA Book separates decision logs from assumptions logs and asks for the reasons, timing and responsibility behind them. It also calls for documenting conditions on which analysis remains valid. The simple record below adapts that distinction; it is not a tested behavioural intervention. Read the AQuA Book's documentation guidance.
A worked decision record
Here is a fictional example. The dates, names and details illustrate the format, not a real client result.
Decision and date: On 4 November, run a four-week trial of the new enquiry process with one service team, rather than introduce it everywhere.
Decision owner: Sam, the service manager. The wider team has provided advice; Sam is responsible for the choice.
What we know: The current process contains a manual handover. The trial team has confirmed it can support a limited test within its existing roster. The process map and roster confirmation are linked in the project workspace.
What we are assuming: The new handover will reduce repeat work. The trial team will encounter enough relevant enquiries for us to assess whether the process is workable. Neither expectation has been established yet.
Options considered: Keep the current process, run a limited trial, or roll out to all teams. We chose the trial because the wider demand and support requirements remain uncertain. Keeping the current process would postpone learning; a full rollout would commit more people before those questions are answered.
Trade-off accepted: Learning from one team may not tell us how the process will work elsewhere. The trial will also need staff attention that could be spent on other improvements.
What would make us reconsider: If the trial cannot operate within the agreed roster, or a required check cannot be completed, pause and review before continuing. Sam will review the evidence with the team at the end of the four weeks. A wider rollout needs a separate decision.
Notice what this does not say: “The trial will work.”
It explains why the team is trying it, what remains uncertain and where its authority ends. Those details remain useful whether the trial succeeds, struggles or produces an inconclusive result.
Leave room for an honest alternative
A decision record can become a defence of the preferred answer if the other options are barely considered.
Instead of “We reviewed the alternatives”, name a credible alternative and the reason for setting it aside. Include waiting or gathering more information when either was genuinely possible.
You don't need an exhaustive list. You do need enough detail to distinguish a considered trade-off from an option nobody examined.
If useful information is spread across the team, gather it before finalising the record. This is where getting independent views before a team decision can help. The record captures the resulting choice; it doesn't replace the conversation that informs it.
Add new information without rewriting the past
Suppose the fictional trial loses access to the staff member who was going to support it.
Add a dated update explaining what changed, what it affects and what you are now deciding. Keep the original entry available.
Otherwise, a record written before the decision can quietly become a polished explanation written afterwards.
Use your organisation's existing version history or record-keeping process. Correct factual errors transparently, and follow applicable privacy and retention requirements. This isn't a reason to retain sensitive information unnecessarily.
When the review date arrives, open the original record before writing the review. Which expectations held? Which failed? What became known only later? Did a warning exist at the time but go unaddressed?
If the evidence supports a different choice, explain it. Changing your mind as a leader is easier for others to follow when they can see what changed and what happens next.
Try it before your next consequential choice
Use these six prompts:
- What are we deciding, when, and who owns the decision?
- What information are we relying on?
- What are we assuming or still unsure about?
- Which realistic alternatives did we consider?
- Why this option, and what trade-off are we accepting?
- What would prompt a review, and who will arrange it?
Then return to the record when there is something to learn.
The useful question is not “Can I make my earlier decision sound sensible?” It's “What did we understand at the time, and what do we understand now?”
Bring a real decision into your leadership session
If your team would benefit from practising how it questions assumptions and explains decisions, explore Andy Nunn's Leadership Mindset Workshop. Share the challenge your leaders are facing to discuss whether a workshop would fit.