The anatomy of a decision memo

A memo that describes a situation is a report. A memo that forces a choice is a decision document, and the two are not the same artefact wearing different clothes. The difference shows up in structure long before it shows up in prose.

What a decision memo owes the reader

The reader of a decision memo is not there to be informed. They are there to decide, usually under time pressure, often having already formed a view. The memo either helps them test that view or it wastes the meeting.

That produces three obligations:

• State the decision in the first paragraph. Not the background, not the methodology. If the reader stops after four sentences they should still know what they are being asked to approve.
• Show the constraint that actually binds. Every situation has a dozen considerations and usually one that decides the outcome.
• Name the option you rejected and why. A memo with a single option is a recommendation with the workings removed.

The recommendation is the easy half. The half that matters is knowing which constraints are load-bearing and which are merely true.

The ordering problem

The instinct is to build to the conclusion, which is how the analysis was actually done. It is the wrong order for the reader, because it asks them to hold every input in their head until the last page. Invert it. See our note on advisory practice for why the widening step happens earlier than most people expect.

A shape that usually works

1. The decision, in one sentence.
2. The recommendation, with the single strongest reason.
3. The binding constraint.
4. What was rejected, and the condition under which that would change.
5. Everything else, which most readers will skip and one will read closely.

Point five matters more than its position suggests. There is always one reader who goes straight to the appendix, and they are usually the one who can kill the recommendation.

Length, and why it is a structural question

Length is treated as a style problem and it is not. A memo runs long because its structure forces the reader to carry context, not because its sentences are wordy. Cutting adjectives from a badly ordered memo produces a shorter badly ordered memo.

The three-page memo is where most of the damage happens, because it is long enough to bury the recommendation and short enough that nobody thinks to add a summary.

---

A note on tooling

None of this is a document-formatting problem, though it gets treated as one. A shared template helps at the margin. Some teams enforce structure in the tooling itself, so a memo without a ## Decision heading simply will not publish, which works better than it should.

The above is a house view, not a rule. The best memo any of us has read broke three of these and was six pages long.