Decision Log: Recording What Was Decided, When and On What Basis

A decision log records each significant decision at the time it is made: what was decided, when, on what reasoning, with what confidence, and what would change the decider's mind. The entry is written before the outcome is known. Memory rewrites reasoning toward whatever happened. Later, it lets a decision be judged by its basis rather than its result, and the line about changing one's mind is the most valuable in it.

People remember having been right. Not because they lie, but because a year after a decision the only thing still vivid is how it turned out, and the reasoning quietly reshapes itself to fit. The only reliable witness to what was actually thought at the time is a note written at the time. A decision logA dated record of decisions, kept by the person or team making them, with the reasoning written down at the moment of deciding rather than reconstructed later. is that note, kept as a habit. One entryOne record in the log: a single decision with its date, reasoning and conditions, written as a unit so it can be read alone later. has six lines, and it must be written before the result is in. The line naming what would change the decider’s mind turns out to be worth more than the line saying how sure they are, and a year later the log is read for its basis rather than as a league table. Ishaan Verma wrote such an entry in January, before recommending a Rs 25,00,00,000 loan to Kaveri Cold Chain Private Limited.

What is a decision log, and what is it defending against?

Start with an argument at a dinner table. In March a household bought a second-hand scooter for Rs 42,000/- rather than a new one for Rs 78,000/-. By December it has eaten Rs 15,000/- in repairs, and now everybody remembers having had doubts. The father recalls saying the engine sounded rough. The daughter recalls saying the seller seemed too eager to be rid of it. Nobody recalls what they all said in March: that the saving would pay for the daughter’s coaching fees. It did. Every person at that table is sincere, and every one is remembering March through December. Nothing written in March survives, so December wins.

A decision log defends against exactly this: the mind's habit of judging a decision by how it turned out and then rewriting the reasoning to match. The habit has a name, hindsightLooking back at a decision after its result is known. Hindsight makes the result feel as though it was predictable at the time, even when it was not., and it is not a character flaw; it is how memory works. Once the outcomeWhat actually happened after the decision: the result, as it stands at the date of review, whether or not it has finished playing out. is known, the mind cannot un-know it, so it reads the past as though the outcome had been obvious. A good decision that went badly starts to look careless; a careless decision that went well starts to look shrewd. Judging by result alone would be fine if the world were fair, but the world hands good outcomes to bad reasoning often enough that a team judging by result learns the wrong lessons and rewards the wrong habits.

The grid below is the shape of the whole argument. Across the top runs the outcome, bad to good. Down the side runs the basisThe reasoning and inputs a decision rested on at the moment it was made: what was known, what was assumed, and why the assumptions were believed., weak to sound. Every decision sits somewhere in the four cells, and a reviewer who has only the outcome knows only which column it is in. Kaveri Cold Chain, twelve months on, sits in a column that is neither clearly bad nor clearly good: the loan was serviced on time, but profit before tax reached Rs 1,10,00,000 against a forecast of Rs 4,20,00,000. Which row it belongs in, sound basis or weak, is a question the outcome cannot answer. Only the entry can.

What does one entry contain?

An entry is short. Six lines, written in ten minutes, on the day. The first three answer the reviewer's factual questions: what exactly was decided, on what date, and on what reasoning. The next three are the ones people skip, and they are the ones that make the entry worth keeping: how sure the decider was, what would change the decider's mind, and when the entry will be reopened. Each line exists for what a reviewer can do with it later, not for its own sake. Look at the figure below and read the right-hand column first.

Six lines: what, when, why, confidence, what would change my mind, and a review date, each written for the reader a year away rather than for the writer today. "What" is the decision in one sentence, specific enough that two people cannot disagree about whether it was taken. "When" is the date, and the date fixes the entry against the flow of information: everything known before that date was available, and nothing after it counts. "Why" is the reasoning, with the inputs it rested on and the source of each. ConfidenceThe decider's own estimate of how likely the decision is to work out, stated as a word or a rough probability at the time of deciding. is an honest word or a rough figure, not a promise. "What would change my mind" is a short list of things that could be observed and that, if observed, would reopen the decision. And the review dateThe date named in the entry when the decision will be looked at again, whatever has happened, so the reopening is planned rather than forced by events. is the appointment with the decider's future self, so the reopening actually happens rather than waiting for a crisis.

Why must the entry be written before the outcome?

One rule makes everything else work: the entry is written before the result is known, or it is not an entry. Not the week after, when the first numbers arrive. Not "as soon as I get a moment". Before. The reason is not tidiness. Once even a hint of the outcome is in, the reasoning written down is no longer what the decider thought at the time; it is what they now think they thought, and the two drift apart in a direction nobody can feel.

The entry is fixed at the decision date, memory drifts toward the outcome, and the review reads the entry precisely because it is the one thing that did not move. Watch the timeline below. Ishaan Verma's entry, written in January, says occupancy will reach 80 per cent. The written line stays flat. Ink does not update. His memory of what he forecast is the dashed line: flat for a while, then bending downward as the monthly occupancy figures come in below plan, until by the twelfth month, with actual occupancy at 71 per cent, he half-remembers having said something nearer 73. He is not lying. Nobody is. But if the entry had been written in month twelve it would have said 73, and the review would have found a forecast that was nearly right and a decision that looked better than it was.

What is "what would change my mind", and why is it the most valuable line?

Ask a street vendor who has just taken a stall outside a new office block how sure she is that it will pay. She will say "quite sure". Ask her instead what would make her give the stall up, and she will say something far more useful: "if the canteen inside reopens, or if I am still under two hundred cups a day by Diwali". The first answer is a feeling. The second is a pair of things anyone can check on a date. The second answer is what the change-my-mind line asks for. In the trade the items on it are sometimes called kill criteriaConditions named in advance which, if they occur, mean the decision should be reopened or reversed. Written so that someone else could check whether they have happened.: the observable, dated conditions under which the decider has agreed, in advance, to reopen the decision.

Someone else can check the change-my-mind line at a dated point, and it records what the decider was watching for. A review needs exactly those two things, and the confidence line supplies neither. Confidence has its uses: over many entries it shows whether a person's "very sure" means anything. But on a single entry, "confidence: high" tells a reviewer almost nothing about the reasoning. "Occupancy below 68 per cent at six months would change my mind" tells the reviewer three things at once: that occupancy was the variable being watched, that 68 was the level judged to break the thesis, and that six months was the horizon. Twelve months later, the reviewer can hold that line up against what happened, and, more valuable still, can see what the line did not name.

The bars below are illustrative weights for scoring an entry, and the builder further down applies them. No standard sets them. But the order is the point. The change-my-mind line carries about a third of an entry’s value, the reasoning a fifth, and the confidence word a tenth. People give them roughly the reverse attention.

How is a log used later without turning it into a scoreboard?

A log that nobody reads is a diary. A log that is read the wrong way is worse. The right way is a review on the date the entry named, with two questions asked in order and the second never before the first. First: given what the entry says was known and reasoned in January, was the basis sound? Second, and only then: what happened, and what does the difference between the entry and the outcome teach? Ask them in that order and the log improves the next decision. Ask them backwards, or ask only the second, and the log becomes a record of who was right.

A testable line is a line that can be scored against. A log read as a scoreboard turns its entries vague within a quarter; a log read for its basis keeps them specific. Watch what happens in the two panels below. On the left the team reads entries at the review date, judges the reasoning, and asks what the change-my-mind line missed. The entries in the second and third quarters are as sharp as the first. On the right the team keeps a tally of whose logged decisions worked, and by the second quarter the change-my-mind lines have softened to "if occupancy disappoints", and by the third to "subject to market conditions", and then they vanish. Nobody decided to do this. Each person simply stopped handing the scorer a stick.

What did Ishaan Verma's Kaveri entry say?

Now the worked entry, exactly as it stood in January. Ishaan Verma, an invented analyst on an investment team, decided to recommend lending Rs 25,00,00,000 to Kaveri Cold Chain Private Limited, a refrigerated-warehousing business. He wrote the entry the day before the credit committee met. The callouts on the right are what a reviewer wrote on the entry twelve months later. Read the facsimile line by line, then read the callouts.

Twelve months on, the entry's greatest value is not that it shows what Ishaan Verma got right or wrong, but that it shows exactly what he was watching for and exactly what he was not. Set the entry beside the outcome. The contract delivered its 14 points, so the first trigger never fired. Occupancy reached 71 per cent at twelve months, above the 68 line, so as far as the record shows the second did not fire either. The tariff held at Rs 1,150 per pallet-month, so nor did the third. Every trigger he named stayed quiet, and profit before tax still came in at Rs 1,10,00,000, a shortfall of Rs 3,10,00,000 against the forecast. Why? Because the two things that did the damage, a competitor opening nearby that took the spillover to zero and a power tariff revision that lifted power cost from Rs 1,80,00,000 to Rs 2,30,00,000, were not on the line at all. The finding is more useful than the outcome itself: not "the forecast was wrong", but "the watch list had two holes, and here they are". The next entry Ishaan Verma writes on a warehousing loan will name competitor capacity and power tariff, and that improvement is the whole return on keeping the log.

How do lenders, analysts and households actually use a decision log?

A credit team uses it as the memo behind the memo. The formal credit note says what the committee approved; the analyst's log entry says what the analyst was watching for and when. When the six-month review comes round, the team does not ask "how is Kaveri doing" in the abstract; it takes out the entry and asks whether any named trigger has fired and whether anything unnamed has appeared. The meeting is much shorter and sharper, and it produces a better next entry.

A research analyst uses it as the record of basis. Where a recommendation goes to clients, the regulator in India expects the reasoning behind it to be recorded and kept, and the log entry is the natural home for that basis, written before the call was published rather than reconstructed afterwards. The idea of a personal decision journal, kept by investors to separate skill from luck over many decisions, is associated with the psychologist Daniel Kahneman and has been widely adopted in investing practice; the log in this guide is that idea with the change-my-mind line given the weight it deserves. Practitioners use the log less to find out whether they were right than to find out what they were not looking at. The blind spot is the part they can fix.

A household uses it at wedding scale. When a couple decides to spend Rs 6,00,000/- on a function rather than Rs 3,00,000/-, a note recording that the bigger hall was chosen because both grandmothers can attend, that the cost is covered from the fixed deposit without a loan, and that the plan will be scaled back if the guest list passes 350 or the caterer's quote rises above Rs 900/- a plate, is a decision log entry. The note will not stop the December argument about whether the money was well spent. The note will settle what was actually thought in March.

Kaveri Cold Chain Private Limited, Ishaan Verma, the investment team, the pharma client and the competitor are invented.
Educational material. Not advice on any investment, tax, budget or market position.