Memory & Context

Why preserve the reasoning behind a decision?

Short answer

A decision without its reasons is an instruction with no manual: you cannot tell when it applies or when it has stopped applying. Keeping the why, the rejected alternatives and the starting conditions lets you judge whether the decision still holds. A six-line template is enough to start.

A thin copper line runs from a grey stone block engraved with fine lines to several glass markers set in a row; an identical block beside it, with no line, stays isolated.

Somewhere in a shared folder there is a line that reads: “Revisions beyond two rounds are billed.” Nobody remembers who wrote it, or why two and not three. A new hire applies it word for word to a small, loyal client, the client is taken aback, and the team spends an hour rebuilding a story that someone knew perfectly well a year ago.

The decision was a good one on the day it was made. What went missing was the why. Here is what to keep alongside a decision, a six-line template for doing it without losing your afternoon, and the cases where it is not worth the trouble.

A decision on its own is an instruction with no manual

A bare decision tells you what to do. It does not tell you in which circumstances, until when, or what else was on the table. The reader is left with two options: obey it literally, or ignore it because it looks arbitrary. Either way nobody is really thinking, because the original thinking is gone.

It gets worse with time. Six months on, the decision circulates as a rule, cut off from the situation that produced it. People apply it to cases it was never meant for, or work around it in the cases it was. And when an AI reads that line, it applies it confidently, with no hint that it comes from a different context.

This is the same gap we described in our piece on AI memory versus context: keeping a piece of information is not the same as keeping what lets you place it.

What the why holds on to

Writing down the why means keeping five things.

  • The reasons. The main motive, stated plainly: “One client kept stretching projects and the extra revisions cost us more than the margin.”
  • The starting conditions. The state of the world when the choice was made: team size, budget, constraints, the tool in use.
  • The rejected alternatives. What you looked at and why you did not take it. This is often the most useful part, because it spares the next person the same round table.
  • The date. A decision with no date cannot be placed against everything that has happened since.
  • What has changed since. This one gets filled in later, but it earns its spot: it is what tells you whether the decision is still valid.

None of these takes much writing. What they take is doing it while you still remember.

An example: the two-rounds rule

Picture a three-person agency. The example is made up, but the mechanism is common.

In March, a large client piles on revision after revision for weeks. The agency decides that from now on, revisions beyond two rounds will be billed. It turns down two other ideas: raising every price (too risky for the small clients) and capping hours per project (too rigid for projects that really do evolve).

By November the agency has six people. The large client left a while ago. A new hire drafts a quote for a small client of many years and applies the rule as written. The client objects, and fairly: the rule was never meant for them.

With the bare line, the team argues. With the why, it sees in thirty seconds that the starting condition (a client who was stretching things) is gone, that the rejected alternatives are still rejected for good reasons, and that rewriting the rule beats either enforcing it or deleting it outright. The conversation is about the right thing.

The why also tells you when a decision expires

People tend to think the why exists to justify the past. Its better use is watching the present. The starting conditions work like an expiry date: as long as they hold, the decision holds. When one of them changes, it is time to reread.

Without that, you only learn a decision is stale by running into it. We have written about what happens when a decision changes: the old one should neither stay as it was nor vanish. A companion piece looks at the signals that a memory of any kind has gone stale, not just decisions (how to tell that a memory has become obsolete). Both start from the same place: written reasons that you can compare with today’s situation.

A decision template in six lines

This is the shortest format that works. Copy it as is into a document, a notes app or an AI memory.

  • What: revisions beyond two rounds are billed.
  • Why: one client kept stretching projects, and the revisions ran past the margin.
  • Rejected alternatives: a general price increase; an hours cap per project.
  • Conditions: team of three, one client involved, fixed-price projects.
  • Date: March.
  • Revisit if: the team doubles, or a long-standing small client is affected.

The last line is the most valuable. It turns the template into a reminder: you do not reread everything, you reread when one of the conditions moves. Write it at the moment of the decision, because that is when you know best what could bring it down.

The same template helps in the opposite situation, when someone wonders whether a question like this was already settled. Finding the note means you do not redo the tradeoff, and you do not hand it over either. That is the subject of our piece on retrieving past tradeoffs without delegating the decision.

Where to keep these notes

For a handful of decisions a year, a shared document or a well-kept notebook does the job. Do not add a tool for three notes.

Things get harder when decisions are numerous, spread over months, and read by several people or several AI assistants. That is where a dedicated memory can help: Nawame keeps decisions, their reasons, the rejected options, constraints and changes over time, so the AI you authorize can use them. It does not replace the template or your judgment. It keeps the template within reach.

One useful caution: writing something down does not make it decided. A written hypothesis is still a hypothesis, and it is better labeled as one. We go through this in the piece on the difference between a memory, a fact, a decision, a hypothesis and an instruction. Practical questions about what Nawame keeps are collected in the FAQ.

Limits of this method

First, the why is often reconstructed after the fact. People write tidy reasons when the decision was partly made out of fatigue, instinct or pressure. A note written three weeks later is less reliable than one written the same day, and both remain someone’s account.

Second, not every decision deserves a note. Picking a button color or the day of the team meeting does not need six lines, and forcing the ritual on everything tends to get it dropped for the things that matter. A decent rule: write one when the decision would be expensive to redo, or when several people will have to live with it.

Finally, a stale note is worse than no note if nobody rereads it. The format helps you know when to reread, but it does not reread for you. And the listed conditions are never complete: there will always be a factor nobody thought of.

One thing to try this week

Pick the last decision that took you real time and forced you to turn down some options. Write its six-line note in ten minutes, without trying to make it pretty. Then ask yourself one question: a year from now, will someone who stumbles on this text understand why you chose this, and know when to question it?

If the answer is yes, do the next decision. If the “Revisit if” line resists you, that usually means the decision rested on an assumption you had not yet put into words. To try it on your own topics, Nawame is free for up to 10 memories.

Your next prompt should not have to contain your whole story.

Start building a memory your AI tools can find again when it matters.

Start free