Write Down Why, Not Just What Was Decided

A short decision record can help a team reconsider a choice without starting the argument from zero

By Arbab Naseebullah Kasi, Chief Executive Officer - CEO, Feel Worldwide Foundation Inc.

A team remembers that it chose a particular approach. What it cannot remember is why.

The choice remains visible in the calendar, the software, or the way work moves between people. The reasoning has disappeared into a conversation nobody can now reconstruct. One person recalls a budget constraint. Another remembers a concern about timing. A third is sure the arrangement was supposed to be temporary.

The next discussion begins with an unnecessary disadvantage: the team must debate the decision before it understands the problem the decision originally solved.

A short decision record can help. Not a transcript of every meeting, and not an elaborate defense of whoever was in charge. Just a clear account of the choice, the circumstances, the trade-off, and what might make the choice worth revisiting.

The purpose is not to preserve every decision forever. It is to preserve enough reasoning to change decisions intelligently.

Borrow a useful habit without importing an entire system

Software teams have a documented practice for this problem: the architecture decision record. In its guidance, Amazon Web Services describes a record that captures a significant technical decision, its context, and its consequences. A collection of these records forms a decision log. AWS architectural decision record process.

That guidance is written for software development. It is not evidence that the same method will improve every nonprofit, small business, or community group. But its central idea is worth adapting: record the reasoning at the point when people still know it.

The adaptation can remain modest. A small team may need a shared document rather than specialist software. A project may need three useful records rather than a record for every choice. The test is whether a future reader can understand what would otherwise require finding the people who attended the original discussion.

Consider decisions that affect how others work: choosing one shared intake process, limiting the scope of a pilot, or assigning responsibility for a recurring service. These are stronger candidates than a routine choice that will have no continuing consequences.

Begin with the problem, not the preferred answer

A weak record begins, “We selected Option A because it was the best option.” That is an announcement wearing the costume of an explanation.

A stronger record names the difficulty the team was trying to resolve. Was information being collected twice? Was a process too slow for the people using it? Was the team trying to learn something before making a larger commitment?

Then it identifies the constraints that mattered at the time. A deadline is different from a preference. A confirmed limitation is different from an assumption. “We have one trained coordinator available” carries a different meaning from “we believe one coordinator will be enough.”

These distinctions make the record useful later. A constraint that has disappeared may justify a new choice. An assumption that proved wrong may require a correction. Without the distinction, both can harden into an unexplained rule.

Avoid reconstructing a polished rationale after the outcome is known. If the record is written later, say that it is a retrospective account and identify what remains uncertain. Honest incompleteness is more useful than a confident explanation nobody actually used.

Show the cost of the choice you made

A decision rarely offers only advantages. It may simplify one person's work while creating extra steps for another. It may allow a quick start while limiting what the team can learn. It may reduce coordination effort while making the service less flexible.

Michael Nygard's original discussion of architecture decision records explicitly includes positive, negative, and neutral consequences. The format is intended to make the reasoning visible to future colleagues, rather than leaving them to accept or reverse an earlier choice without understanding it. Documenting Architecture Decisions.

For a nontechnical team, the practical adaptation is straightforward: name the disadvantage you are accepting, not just the advantage you hope to gain.

That is not an invitation to write a defensive list of every imaginable risk. Choose the trade-off that would matter to someone implementing or reviewing the decision. If the cost falls on another person or group, make sure their perspective has been heard rather than describing the inconvenience only from the decision-maker's chair.

A record does not make an unfair choice fair. It makes the reasoning available for scrutiny.

An example that can survive the next meeting

Imagine a small organization deciding how to receive requests for a new community workshop series. The following example is fictional and illustrates a working method, not an actual program or result.

The team considers two approaches. It could let each workshop facilitator receive requests independently, or it could route requests through one shared contact point. Independent handling may be faster for familiar participants. A shared contact point may make it easier to notice duplicate requests and allocate places consistently.

A useful record might begin with a direct decision: “For the first workshop cycle, requests will come through one shared contact point, with an assisted route for people who cannot use the standard form.”

The context would explain that the group is testing a new series and needs a reliable overview of requests. The rationale would say that consistent allocation is the priority for this cycle. The accepted drawback would name the additional coordination step and the possibility that replies take longer.

The record should also identify what is not being decided. It does not mean every future program must use the same route. It does not authorize collecting unnecessary personal information. It does not mean facilitators should ignore someone who approaches them for help.

Finally, it would identify who is responsible for implementing the arrangement and what information should be reviewed before the next cycle. For example: where did requests stall, how often was assistance needed, and did the shared route actually resolve duplicate handling?

None of these sentences requires a long policy document. Together, they prevent a temporary operational choice from becoming a permanent rule by accident.

A draft is not an approved decision

A clear explanation can still create confusion if readers cannot tell its status.

Label a proposal as a proposal. Identify an accepted decision as accepted, with its date and the person or body that had authority to make it. If an option was rejected, do not leave it looking like a live instruction.

This is especially important when a document is shared before discussion. A carefully written recommendation can look final simply because it appears finished. A participant may begin implementing it while another still believes the group is considering alternatives.

For everyday project coordination, a small status line can prevent that misunderstanding. For decisions governed by formal approval requirements, a convenient decision note does not replace the required process or official record.

The discipline is to represent authority accurately. Writing a decision down does not create permission to make it.

Revisit the assumption, not the argument

A record should help a team notice when the reason for a choice has changed.

In the workshop example, the shared contact point might be appropriate during the pilot but become a bottleneck as the program grows. Alternatively, the first cycle might show that it works well and needs only clearer responsibilities. The review should examine that evidence rather than replaying the original preferences as if nothing had been learned.

A helpful question is: “Which part of our reasoning still holds, and which part no longer does?”

That question creates a different conversation from “Who was right?” It allows a decision to have been reasonable under earlier conditions and no longer suitable under current ones. It also leaves room to acknowledge that an earlier assumption was weak without pretending the problem was inevitable.

When a choice changes, preserve the connection between the old and new decisions. AWS's best-practice guidance recommends retaining superseded records and linking the history, rather than erasing earlier reasoning. It also recommends a central location where the relevant team can find the records. AWS decision-record best practices.

For a small organization, that might mean marking the old note “replaced” and linking to the new one. The old record explains the past; the new one governs the current arrangement.

Keep the record usable, proportionate, and respectful

More detail is not always more helpful. A record overloaded with meeting dialogue, personal criticism, and unrelated background becomes harder to use and harder to share appropriately.

Write for the colleague who needs to understand the choice, not for an imagined audience judging every sentence. Link to supporting material when it belongs elsewhere. Include only the personal information genuinely needed for the purpose, and use the organization's appropriate access and retention arrangements.

A practical starting point is a short note with a descriptive title, date, status, context, decision, rationale, accepted drawbacks, and review trigger. This is a suggested working format, not a universal standard. Remove fields that do not help; add one when a real need appears.

Try it on the next consequential but manageable project decision. Ask someone who missed the discussion to read the note. Can they explain what was chosen, why it was chosen, and what would justify reconsidering it? Their confusion is useful feedback about the record.

A team does not need perfect institutional memory. It needs enough shared memory to avoid treating yesterday's compromise as an unexplained law, or tomorrow's revision as a betrayal.

Which decision in your team is still shaping the work, even though its original reasoning is becoming difficult to find?

Next
Next

Before Choosing a Career, Watch an Ordinary Workday