Write the one-pager

You want a decision. A migration, a rewrite of the sync layer, two weeks to fix the thing that pages you on Sundays. You have thought about it more carefully than anyone else in the company and you are fairly sure you are right.

So you put it on the agenda, explain it well, and get let’s revisit next quarter.

Writing is not the persuasive medium

Before recommending the one-pager I have to report the evidence that cuts against it, because it is better evidence than anything supporting the usual advice.

Schroeder and Epley ran four experiments where evaluators watched, listened to, or read the same pitch. The finding: evaluators “rated a candidate as more competent, thoughtful, and intelligent when they heard a pitch rather than read it.” Having actors read the written pitches aloud produced the same effect, so it is the voice doing the work, not the words. A separate 2017 study of over a thousand participants found slide decks and purely spoken presentations scored the same on comprehension and recall.

There is no measured basis for the claim that written proposals win more often. If your goal is to seem impressive in the room, talk.

Two things a meeting cannot do

Write it anyway, for reasons that have nothing to do with impressiveness.

The argument has to survive rooms you are not in. Your director will re-explain your proposal to their peers on Thursday, without you. What travels is whatever they retained. A document is the only version of your reasoning that arrives intact when you are not there to defend it.

Writing exposes what you had not thought through. This is the part Amazon actually claims, and they have never claimed the other thing. From Bezos’s 2004 email banning slides: “Well structured, narrative text is what we’re after rather than just text. If someone builds a list of bullet points in word, that would be just as bad as powerpoint.” Bullets let you skip the connective tissue. Prose does not, which is why writing it is uncomfortable and why the discomfort is the useful part.

Their 2017 shareholder letter is blunt about the cost: people “mistakenly believe a high-standards, six-page memo can be written in one or two days or even a few hours, when really it might take a week or more.” And the meeting itself is silent, because they read it there: “We silently read one at the beginning of each meeting in a kind of ‘study hall.’”

The move

Write the page you would want if somebody else were proposing this, then hand it over before the meeting rather than presenting it during one.

What goes on it

Shape Up’s pitch format is the best short-form skeleton in circulation, and its five parts are almost exactly the five questions a sceptical reader asks. Problem, the raw idea or situation motivating the work. Appetite, how much time you want to spend and how that constrains what you build. Solution, the core elements in a form somebody can grasp immediately. Rabbit holes, the details worth calling out so nobody falls in. And no-gos, what you are deliberately not doing.

That fourth and fifth are what separate a proposal from a wish. Anyone can describe an outcome. Only somebody who has genuinely thought about it can tell you where it gets dangerous and what they are refusing to include.

Keep it short on purpose. Amazon’s PR/FAQ specification is one page for the press release and five or fewer for the questions, with a line worth pinning above your desk: “There are no awards for extra pages or more words.” They call the length limit a forcing function, and that is right. The constraint is not politeness to the reader. It is what makes you decide what actually matters.

BEFORE                              AFTER

"Quick thing on the agenda:         PROBLEM     sync fails silently ~40x/week.
 I think we need to rewrite                     support finds out before we do.
 the sync layer"
                                    APPETITE    two weeks. not a rewrite.
  - explained live, once
  - survives as whatever your       SOLUTION    idempotency keys + a dead-letter
    director remembers                          queue + one alert. sketch below.
  - no record of what you
    ruled out                       RABBIT      the legacy partner API retries
  - the risky parts never  HOLES                without a key. needs a shim.
    came up
                                    NO-GOS      not touching the scheduler.
                                                not migrating the old records.

                                    sent Tuesday. meeting Thursday.

The 2026 problem

Here is what is different about writing this now, and it is the reason to write it by hand.

Producing a plausible one-pager is free. Everyone knows that, including your reader, and the research on how they respond is not comfortable. Across four preregistered experiments with 4,439 participants, people who used AI at work “anticipate and receive negative evaluations regarding their competence and motivation”, and observers rated them “lazier, less competent, and less diligent than people who get help from other sources.”

So the artifact is now cheap and the credibility is not. Which means the sections that carry weight are precisely the ones a model cannot fill in for you: the appetite, because it is a commitment you are personally making; the rabbit holes, because they come from knowing this codebase; and the no-gos, because they are you choosing to be constrained.

Use whatever you like to tidy the prose. Do not let anything else write the constraints, because the constraints are the entire signal.

Try this week

Take the thing you have argued for twice and lost, and give it one page.

Write the no-gos section first. Not the solution, the no-gos. It takes ten minutes and it is diagnostic: if you cannot name what you are refusing to build, you do not have a proposal yet, you have an enthusiasm, and that is probably why it lost.

Then send it the day before the meeting, with one line: This is what I’d like to decide on Thursday. Happy to be told no, but I’d like the no to be about the appetite rather than the idea.

You will still be more persuasive out loud. But out loud does not survive Thursday.