Decision logs: writing down why you chose what you chose

You will remember the decision. You chose the vendor, the pricing model, the software, the process. What disappears is the reasoning — which alternatives were actually on the table, what mattered most at the time, and what you knew when you decided.

This matters more than it sounds, because a decision without its reasoning cannot be reviewed. When someone asks why the business uses this supplier, the available answers are 'we compared them and this one won' and 'I do not remember'. Neither tells you whether the comparison still holds. If the deciding factor was delivery speed and delivery speed has stopped mattering, the decision should probably be revisited; if it was a compliance requirement, it should not. Without the record, both cases look the same and get treated the same way — usually by leaving it alone, because changing something for unknown reasons feels riskier than keeping it.

The expensive version happens when a new person arrives, sees an arrangement that looks obviously suboptimal, and either changes it into a problem that was already solved or leaves it alone while assuming everyone before them was careless.

Six fields, and it should fit on one row of a spreadsheet.

Date. Not for tidiness — it establishes what information was available, which is the whole basis for judging the decision later.

The decision, in one sentence, stated as what was chosen rather than what was discussed.

The options considered, including the ones dismissed early. This is the field people skip and the one that saves the most time, because it stops the business from re-evaluating an option that was already ruled out for a reason.

The criteria, ranked if possible. What actually mattered: cost, speed, risk, reversibility, who has to use it every day.

The reasoning: why the chosen option won against those criteria. Two or three sentences.

What would change your mind — the specific condition under which this decision should be reopened.

An outcome field, filled in later, closes the loop. It is the only field that tells you anything about the quality of your decision-making rather than the decision.

'Chose vendor B' is a record. 'Chose vendor B because their lead time was two days shorter and the team using it daily preferred the interface, accepting a higher unit cost' is a decision that can be reviewed by someone who was not in the room.

What makes reasoning useful is naming the trade-off you accepted. Almost every real decision gives something up, and the thing given up is what you need to know later — because that is where the problem will surface. A business that recorded 'accepted a higher unit cost' can look at a margin complaint six months on and recognise it as the known price of a deliberate choice rather than a mystery. Without the record, the same complaint reads as a discovery.

This also protects against a subtler failure: quietly rewriting your own reasoning after the fact. Outcomes are noisy, and a decision that turned out badly is remembered as having been obviously wrong, while one that turned out well is remembered as having been obviously right. Written reasoning is the only defence against that, and it is the difference between learning from a decision and building a story about it.

This field is unusual and it is the most valuable one, because it converts a decision from permanent into conditional and specifies the trigger in advance.

It has to be concrete enough to notice. 'If they become unreliable' cannot be observed. 'If we get two late deliveries in a quarter' can. 'If our monthly volume passes the point where the tier pricing stops making sense' can. The test is whether someone who was not part of the decision could tell whether the condition has occurred.

Stating it in advance is also the honest way to handle a decision you are not confident about. Plenty of small-business decisions are made on thin information under time pressure, and pretending otherwise in the record is a mistake. Writing 'chose this because we needed something working by Friday and it was the only option that could be running in two days; revisit before the annual renewal' is an accurate account of a reasonable decision. It stops the choice hardening into policy by default, which is how a business ends up with an architecture nobody chose.

It also removes the sunk-cost argument, because the condition for reopening was agreed before anyone was invested in being right.

Three triggers, and only three, or the log becomes a source of recurring work.

The stated condition occurs. This is the whole point of the previous section — the trigger fires, the decision is reopened, and nobody has to argue about whether it is time.

A renewal or commitment point arrives. Contracts, subscriptions and supplier agreements have natural moments where changing is cheap, and those are the moments to read the original entry. Reviewing a decision when switching is expensive mostly produces frustration.

Something the reasoning depended on has changed. If the deciding criterion was a constraint the business no longer has — a headcount, a location, a volume — the decision is running on obsolete inputs even though nothing has gone wrong.

What is not a trigger: a bad week, or someone new disliking the arrangement. Both feel like signals and neither is, and a log that gets reopened on either becomes a debating record rather than a decision record. The value of writing down the trigger is precisely that it lets you decline to reopen without having to defend the original choice again.

It cannot make decisions better on its own. Recording reasoning creates mild pressure to have some, which is a real benefit and a small one. A log full of well-documented poor decisions is entirely possible, and reads as impressively organised.

It cannot settle whether a decision was correct. Outcomes have too much noise: a good decision can produce a bad result and a reckless one can be rescued by luck. What the log supports is judging the decision against the information available at the time, which is the only fair standard and is not the same as judging it by how it turned out. Logs used to assign blame after bad outcomes stop being honest within about two entries.

And it cannot cover everything. A log attempting to capture every choice becomes an administrative burden and then stops. The decisions worth an entry are the ones that are expensive to reverse, that commit the business for a period, or that a future person would look at and wonder about. That is a handful a quarter in most small businesses, not a handful a week — and the discipline to write only those is what keeps the habit alive long enough to pay out.