Definition
A decision log is a running record of decisions, kept in one place for a programme, with one entry per decision. Each entry holds at least the date, the decision, the evidence behind it, the expected effect and the date the effect will be read; the actual effect is added at that later review. Fuller versions add who decided, how easily the decision can be undone, and the guardrail that would reverse it early.
It is a record of reasoning, not of activity. A list of edits says what happened to the settings; a decision log says why the team chose it and what they thought it would achieve, which is the part people forget first and need most when results move.
Decision log and change history
Advertising platforms keep their own change history. The tutorial on auditing change history records, from the platform's own help, what that history holds: who made each edit, when, and often the previous value. It does not hold why an edit was made or what the person expected, and it never sees changes outside the account, such as a site release, a price change or a competitor's arrival.
The decision log fills both gaps. It holds the reason and the expectation, and because it is kept by people rather than by one platform, it can cover every channel and the outside events that affect them. The two are read together: change history for what changed and when, the decision log for why.
In practice: owners, monthly reviews and post-mortems
A log works when one person owns it and reviews read it. The guide to performance reporting places it at the end of the monthly report, so its latest entries are the first thing the next review checks; that is what makes reporting cumulative rather than a series of separate meetings. The marketing decision log template gives the columns and a filled example, and the format belongs there.
Busy periods make the log more important, not less. The peak season trading playbook gives the trading lead the log during the peak window, updated at the end of each daily stand-up. And when a crisis has passed, the tutorial on leading marketing through a business crisis reviews marketing's part using the decisions logged at the time, rather than relying on memory of a stressful week.
A hot sauce brand nearly restarts a campaign it already tested
Glouffell Hot Sauce hired a new marketer in March. Reviewing the paid social account, she found a paused campaign aimed at new customers and planned to restart it with £1,200 a month. The decision log had an entry from the previous October: the campaign had run as a four-week test against a holdout region, spent £2,600, and produced orders worth £1,900 at a 38% gross margin, so it had lost money on first orders. The entry named who decided to pause it, the evidence and a note that repeat purchases from those customers would be checked in January; a second entry recorded that they had not changed the result. She left the campaign paused, added a row recording her decision not to restart it, and put the budget into a test the log showed nobody had run.
Related terms
A Change LogA dated record of changes made to tracking, accounts, reports, lists or rules, with who made each and why, so later movements in the figures can be traced. is the closest relative: a dated record of changes to tracking, accounts, reports or rules, kept so later movements in the figures can be traced, while a decision log records the choice and its expected result. Many entries begin as a HypothesisA testable statement of the change to make, the effect expected on a named metric and the evidence-based reason for expecting it., a testable statement of what a change should do. The expected effect is usually stated in terms of a KPIA metric tied directly to an objective., and it is read against a BaselineAn estimate of what a metric would have done without a given change, campaign or promotion, against which the effect is measured., the estimate of what would have happened without the change.
Common confusions
- The platform's change history is a decision log. It records what changed and who changed it, not why or what was expected, and it misses everything outside the account.
- Only big decisions need logging. Small changes made for a reason, such as a paused audience or a raised target, are the ones most often forgotten and repeated.
- The log is finished when the decision is recorded. An entry without its actual effect, read on the date it set, teaches nothing.
- One log per channel is tidier. Effects cross channels and the review that reads them is usually one meeting, so one log per programme is easier to use.
- A decision log is the same as a change log. They overlap, but a change log traces changes to data and settings, while a decision log keeps the reasoning and the expectation behind decisions.
