Decision Log

What this tool does

CM8-295 is a decision log: the record of what was decided, by whom, on what basis, and what it closed off. One row per decision — the thing decided, the date, whoever held the authority, the options rejected, the reasoning, the damage if it is wrong, how hard it would be to undo, and the assumptions it rests on. The tool tracks the decisions that are effectively permanent, flags reviews that have come and gone, and prints a report someone who was not there can read without being briefed.

Everything runs inside this single file — no account, no upload, no network request — so sensitive reasoning about suppliers, costs and scope stays on your computer.

A decision log is not meeting minutes

Minutes record that a discussion happened. A decision log records what is now true.

Minutes are chronological and tied to a meeting: attendance, the debate, actions. That is a real job, and the Meeting Minutes Manager does it properly. But minutes are close to unsearchable eighteen months later. To find why the archive cut-off is three years you must remember roughly when it was discussed, find that meeting and read the item — and if two people settled it in a corridor and merely reported it, it is not in there at all.

A decision log is indexed by the decision, not the date of a meeting: one row, one decision, standing on its own, whoever took it and wherever. Run both — minutes for what happened, this for what is now the case.

Write the decision as a statement, not a question

"Vendor selection" is a topic. "Which vendor?" is a question. "Appoint Vendor B as the warehouse system supplier on a five-year term" is a decision. Only the third is any use read cold, because only the third says what is true now. The test: can the sentence be acted on out of context? One sentence; the reasoning goes in the rationale.

Options considered: the most valuable column here

Most logs record what was chosen. The good ones record what was rejected and why, and that field does more work than the rest of the row.

It stops the same debate being reopened every quarter: somebody joins, sees the vendor and asks why not the cheaper one, and without the field you re-run the analysis. It is also the first thing an auditor or a new starter asks for — nobody experienced asks "why did you do this?", they ask "what else did you look at?" A decision with no rejected options reads as one that was never really made.

And it checks your own thinking. If you cannot name two options you seriously considered and why they lost, you have ratified a preference rather than decided anything. The tool will not save a one-way decision without this field.

Rationale that outlives the people who wrote it

Write it for a stranger reading in two years, when everyone involved has moved on. Name the constraint that actually drove it — the date, the budget, the skill you did not have — not the adjectives. "Best fit" is not a rationale. "Met every must-have, support within four hours of both depots, and the only option deliverable before the current system leaves support" is one, and it stays useful even to a reader who disagrees. The tool requires a rationale on any high-impact decision, because that is the one which will be questioned hardest.

One-way and two-way doors

This is the single most useful idea in the tool. Some decisions are doors you can walk back through: try it, and if it is wrong, undo it next week at the cost of a little embarrassment. Others are one-way — going back means unwinding a contract, rebuilding something, or accepting that data is simply gone.

The mistake nearly every organisation makes is applying one process to both. Reversible decisions get weeks of deliberation and three meetings, when the cheapest way to find the answer was to try it. Meanwhile the genuinely irreversible ones — which supplier, which platform, what history to migrate before the old system is switched off — are made at the same speed as everything else, because they arrived on the same agenda and got the same slot. Classify honestly, then behave differently. Easy to reverse: decide fast, at the lowest sensible level, and let the result teach you. Reversible at a cost: decide properly, knowing roughly what the undo would cost. One-way: slow down, write the options out, and get someone who disagrees to read it. The impact-against-reversibility chart shows whether your log matches that logic.

Beware false one-way doors: "we could never change this" is often "changing this would be annoying". Reserve the label for decisions where reversal is genuinely off the table, or it stops meaning anything.

Assumptions and review dates

Most decisions are not right or wrong. They are right until something stops being true: the vendor decision holds while the regional support team exists, the scanner decision while the current operating system is certified.

So write the assumption down as a sentence and set a review date. The review date is not a deadline and not a promise to change anything — it is a calendar entry that makes somebody check whether the sentence is still true, which is the difference between catching a stale decision on purpose and discovering it during an incident. When an assumption fails, do not quietly edit the row: set the status to lapsed, which says plainly that this was reasonable, the ground moved, and it needs retaking. The tool flags active decisions past their review date, and separately those high-impact decisions never given one — the governance gap.

Supersede, never overwrite

When a decision is replaced, resist editing it. Editing destroys the evidence of how your understanding changed and makes the log dishonest — it reads as though you always thought the current thing. Set the old row to superseded, name the reference that replaced it, and write the new decision as its own row. Kept that way the log reads as a history: single cutover weekend, then depot by depot once the vendor's staffing became clear. The tool will not accept a superseded row without a named successor.

Reversed is different: undone because it was wrong or circumstances changed, whether or not something replaced it. Both stay. A log with no superseded or reversed entries is usually one that stopped being maintained.

Who decided, and recording authority honestly

"Decided by" is the person or forum that actually held the authority — not whoever typed the row, and not whoever was in favour. Write the honest answer even when it is awkward. The chart of decisions by decider repays reading: one name against nearly everything is a bottleneck and usually why decisions are slow; a long list of one-offs means nobody is clear who may decide what. And if the one-way doors are decided well below where the consequences land, that is worth a conversation.

Where this sits next to a RAID log

A RAID log holds risks, assumptions, issues and dependencies — things still open and being managed. A decision log holds things that are closed: settled questions and their reasoning. They meet at assumptions, and traffic runs both ways. An assumption here that starts looking shaky becomes a risk there; a risk resolved by somebody choosing something produces a row here. Keep both, and keep them apart.

The numbers this tool produces

The reversal rate is the figure people will treat as a score, so read it carefully. Its denominator is the decisions that were live long enough to be followed through; lapsed ones are excluded, because they failed when the world changed rather than because the decision was poor. A very low rate can simply mean nobody will admit a decision was wrong. A high rate on reversible decisions is often healthy; on one-way doors it is the expensive kind.

The spreadsheet workflow

Decisions often start life scattered across notes and email, and retyping them is wasted effort.

• Spreadsheet template saves a CSV whose headings are exactly this tool's column names, with a guidance row giving the date format and the accepted values for the type, impact, reversibility and status lists.
• Fill it in, one row per decision, then delete the guidance row and save as CSV.
• Import spreadsheet reads it back. Columns match by heading, so order does not matter and extra columns are ignored. Rows failing validation — a one-way decision with no options recorded — are skipped and reported by row number.

The file is read in your browser: nothing is uploaded, and importing adds to what is here.

FAQ

How big does something have to be to belong here? If reversing it would cost real time or money, or if somebody might reasonably ask "why did we do it that way?" in a year, log it. Forty decisions that matter beat four hundred that do not, because people will read the first log.

What if we cannot agree who decided it? That is a finding, not a formatting problem. If three people each think somebody else decided, the decision has no owner and will not be defended when it is challenged.

Can a decision be un-lapsed? No. Take it again as a new row, with the current facts and a reference back — the circumstances that made the first version reasonable are gone. Match the review interval to the assumption rather than the calendar; the setting is only a prompt, and the tool never fills a review date in for you.

Saving your work

Decisions, settings and the report header are written to this browser's local storage as you type. That storage belongs to one browser on one computer: another browser, a private window, a second machine or a clean-up tool that clears site data will not have it.

Treat Export .json as the real save — one file containing everything, which Import .json restores anywhere. Export CSV gives you the register for spreadsheet work, including every filtered record rather than only those drawn on screen. Reset asks twice, then erases everything this tool has stored. There is no undo.

Accuracy & disclaimer

The arithmetic here is counts and dates, and the tool does it faithfully. What it cannot do is judge the decisions. A confident rationale and a hollow one look identical in a text field; an option list can be written after the fact to justify a choice already made; a decision never entered is invisible to every figure on the dashboard.

The log records what was decided. Whether the decision was any good is answered at the review date, against the assumptions you were honest enough to write down at the time — not by the entry. This is a record-keeping and governance aid, not project-management, legal or contractual advice.