2026-05-22
May 22, 2026
The deck is theater. A narrative read in silence is the only meeting format that reliably raises decision quality — and AI is quietly about to wreck it.
Watch what actually happens in your next big decision meeting. Someone clicks through forty slides. The presenter controls the pace, the framing, and — through sheer fluency — the conclusion. Half the room skimmed the pre-read on the way in; the other half didn't open it. Questions cluster on slide 6 because that's where the presenter let the room breathe. A decision gets made, but it's made on the most charismatic argument, not the most complete one.
There is a known fix, and it is almost twenty years old. In 2004 Jeff Bezos banned PowerPoint from Amazon's leadership meetings and replaced it with a written narrative — a six-page memo, in full sentences — read silently by everyone in the room for the first fifteen to twenty minutes. No pre-read assumed. No presenter performing. Just the document and a room full of people consuming it at reading speed, which is roughly three times faster than anyone can present it.
The format is having a quiet renaissance in 2026, and not for nostalgic reasons. The thing that makes it valuable is also the thing that large language models are about to take away — so it's worth understanding exactly why it works before you let a chatbot generate your next strategy memo.
Last Friday this newsletter argued for killing the standing meeting — separating broadcast, discussion, and decision into different channels so you stop paying for a room you don't need. Today's argument is the inverse: for the decisions that genuinely require a room, the single highest-leverage upgrade is changing what people do in the first twenty minutes. Stop presenting. Start reading.
The case rests on four mechanics, and none of them is about etiquette.
Reading beats listening on throughput. A competent reader moves at 300+ words per minute; a presenter narrating slides delivers maybe 130. A silent room therefore absorbs two to three times more substance per minute than a presented room — and absorbs it without the lossy compression that bullet points impose. Bezos's own framing in the 2017 shareholder letter is the sharpest version: "The narrative structure of a good memo forces better thought and better understanding of what's more important than what, and how, things are related." Bullets let you gesture at an idea. Full sentences make you commit to it, and commitment is where the logic gaps show.
The study hall solves the pre-read problem permanently. Every program leader has run the meeting where the doc was sent Friday and nobody read it. You can nag, or you can stop pretending. Reading the narrative in the room guarantees a fully-prepared audience every time, with zero dependency on anyone's discipline the night before. That single move converts "a meeting where we get aligned" into "a meeting where we decide," because alignment happened in the silence.
It levels the room. When the frame arrives as a written artifact rather than a live performance, the loudest voice no longer sets the terms of debate. The remote attendee on a laptop and the senior VP at the table read the same words. Junior contributors who'd never interrupt a polished presentation will annotate a paragraph. For distributed programs — which is most programs now — the memo is the equalizer, because it doesn't reward presence in the room.
It produces a durable decision record for free. The narrative, plus the comments and the decision appended to it, is the artifact. Six months later when someone asks "why did we choose this?", the reasoning is searchable, in context, in the author's own words. A deck answers no future questions; a narrative answers most of them.
For program leaders specifically, there's a fifth payoff the original Amazon framing undersells: the narrative scales across teams that can never all be in one room. A program touching eight teams cannot align in a meeting — there's no meeting big enough, and the timezones don't overlap. But a single canonical narrative can travel. Each team reads the same words, comments in the same margins, and the decision-owner resolves the threads. The memo becomes the program's source of truth, and the "meeting" becomes asynchronous by default — which is what most cross-team coordination actually needs. The live room, when you convene one, is reserved for the genuinely contested calls the comments couldn't settle.
There's a cost, and it's the point. A good six-pager is genuinely hard to write — Bezos estimated a great one could take a week. That difficulty is a feature: it front-loads the thinking that a deck lets you defer until you're standing in front of people. The pain of the blank page is the system working.
Which is exactly why 2026's most tempting mistake is letting AI write the memo. An LLM will produce six fluent, well-structured pages in ninety seconds. But the value of the narrative was never the prose — it was the thinking the prose forced out of the author. Automate the writing and you've automated away the forcing function, leaving a document that reads beautifully and was understood by no one, least of all the person who "wrote" it. The high-leverage use of AI here is the opposite: write the narrative yourself, then point a model at it as an adversary — find my unstated assumptions, argue the strongest counter-case, list what a skeptical CFO would ask. Use the machine to pressure-test the thinking, never to skip it.
Try this week. Take your highest-stakes recurring decision meeting and ban the deck for one session. Require a two-page narrative instead. Open with fifteen minutes of silent reading, then discuss. Name one accountable decision-owner in the doc's header before the meeting starts. Watch how much of the usual "let's take it offline" evaporates.
What it is. A decision meeting that opens with all attendees silently reading a pre-written prose narrative (Amazon's canonical form is six pages), followed by structured discussion of that document. It replaces slide presentations for any meeting whose job is to decide, not merely inform.
When to use it. High-stakes decisions where being wrong is expensive: roadmap bets, architecture direction, build-vs-buy, reorg proposals, go/no-go reviews. Especially valuable when the room is distributed, when stakeholders outrank the author, or when past meetings on the topic ended in vague "alignment" with no decision.
How to run it:
When NOT to use it. Pure information broadcasts, quick status syncs, or genuinely two-way brainstorms where divergence beats a pre-formed recommendation. Writing a six-pager for a low-stakes call is overhead theater in the other direction.
Example: a staff TPM facing a contentious platform-migration decision swaps the usual review deck for a four-page narrative, opens with fifteen minutes of silence, and gets a clean go decision with a named owner in forty-five minutes — after three prior meetings that produced only "let's regroup."
Amazon Narratives — Memos, Working Backwards, and More — a16z's conversation on the mechanics behind the six-pager and the PR/FAQ. The most useful external articulation of why the narrative changes decision quality, not just how Amazon formats its docs — the right primer before you try to install the practice on a team that's never seen it.
Build an async culture that actually sticks — LeadDev on why "write it down" initiatives so often collapse into doc graveyards, and what makes the write-first default actually hold. Pair it with the narrative-meeting format: the memo is worthless if your culture doesn't read and act on what's written.
Why Do More Meetings Lead to Fewer Decisions in Enterprise Teams? — A clean framing of meeting overload as a decision-ownership problem, not a calendar problem: meetings slow decisions when they replace a named owner with a diffuse group. The fix dovetails with the narrative format — every memo should name exactly one accountable owner in its header.
"The narrative structure of a good memo forces better thought and better understanding of what's more important than what, and how, things are related."
— Jeff Bezos, 2017 Letter to Shareholders
Don't miss what's next. Subscribe to Critical Path: