The first ninety seconds of a keynote decide whether the next twenty-eight minutes will be received or endured. This is the rule that turns mediocre speakers into great ones, and the rule that almost every speech-writing book buries in chapter four behind 80 pages of structural theory. The room makes its decision about you, the speech, and whether to pay attention before you finish your second sentence. Lose those ninety seconds and the rest is a recovery operation. Win them and the speech is functionally over before it has begun. The audience is now leaning in, waiting for you to finish the argument you have already started.
This is why the 90-Second Rule flips the standard writing process. Most speakers write a keynote front-to-back, starting with the opening hook and working toward the closing line. That sequence is backward. The right sequence writes the central argument first, then the closing, then the body anchors, and then the opening. The opening is written last because the opening has to do a job that depends on knowing what the rest of the speech is going to do. Writing the opening first produces an opening that is loosely connected to the speech, which is what produces the all-too-familiar moment where a speaker delivers a strong first paragraph and then visibly transitions into a different talk for the next twenty-six minutes. The audience feels the seam. The 90-Second Rule eliminates the seam.
What the opening 90 seconds have to do
The opening has four jobs to complete inside the ninety-second window, and the order matters. Job one is to land a specific concrete sentence that gives the audience an immediate reason to keep listening. Not a statistic. Not a setup. A piece of language that creates an information gap the audience now wants closed. Job two is to declare the argument of the speech in one sentence, so the audience knows what claim you are about to defend. Job three is to set the stakes for the audience specifically, so they know why this argument matters to them in particular. Job four is to make a promise about what they will be able to do, think, or decide differently by the end of the talk.
Four jobs in ninety seconds is about 230 to 250 words. That is roughly the first slide of any speech you have ever seen, and most of the time those 250 words contain almost none of the four jobs. They contain instead an introductory anecdote that the speaker likes but the audience does not need, a thank-you to the host, a gratitude paragraph about being here today, and a slow walk-up to the actual argument that begins around minute three. Three minutes is the threshold beyond which a non-trivial portion of the room has tuned out and will not return. The 90-Second Rule says you cannot afford the walk-up. The walk-up is the speech.
The Anchor Method: structuring the body
After the opening is locked, the body is structured around what I call the Anchor Method. The speech is built from ten to twelve anchor points, each one a single sentence that compresses one piece of the argument. The anchors are the spine. Between the anchors is connective material: examples, evidence, stories, rhetorical pivots. The anchors are memorized verbatim. The connective material is internalized as ideas, not language.
This is the durable structure great speakers use. Each anchor is the kind of sentence that ends up tweeted, screenshotted, or repeated by the audience to colleagues over the next week. “If you can’t write the headline in one sentence, the story is not ready to be assigned.” “The audience metric is downstream of the job.” “The book that gets press is the book whose author treated launch as a project, not a celebration.” Anchors are the parts that survive the moment you stop speaking. The rest of the speech is the support architecture that delivers each anchor with enough context that it lands.
The reason the Anchor Method beats both full-memorization and bullet-point delivery is that it gives the speaker resilience. When the room reacts unexpectedly, the speaker can flex the connective material around the anchors without losing the structure. When a memorized speech stumbles, the whole talk loses confidence. When a bulleted speech wanders, the audience cannot tell where the speaker is going. Anchors keep both speaker and audience oriented.
How to find your real argument
The hardest part of writing a keynote is finding the argument you actually want to make. Most speakers write talks about topics. A topic is not an argument. “The future of remote work” is a topic. “Remote-first companies will outperform hybrid companies on retention by 2030 because the costs of hybrid coordination compound silently” is an argument. The difference is that an argument can be wrong, can be defended, and can be remembered.
The argument has to be falsifiable, specific, and consequential. Falsifiable means a reasonable counter-position exists. If no thoughtful person could disagree with you, the argument is too soft to carry a speech. Specific means the argument is bounded. It applies to a particular field, a particular timeframe, a particular kind of actor. Consequential means the argument changes what someone in the audience will do, think, or decide. If the audience can leave the room and behave exactly the same, you delivered a TED-Talk-shaped lecture, not a keynote.
The argument is found through writing, not thinking. Sit with a blank page and write seven possible argument sentences for the speech. Most will be too soft. Two or three will have heat. Pick the one that scares you a little because you can imagine someone disagreeing publicly. That is the argument. Build the speech around it.
The stories that work and the stories that do not
Personal stories are the most over-used and under-engineered element in modern speaking. The speaker who tells a meaningful story about a near-failure with the named outcome is gripping. The speaker who tells five biographical stories with no through-line is unbearable. The test for whether a story belongs in a keynote is simple: does it serve as evidence for one of the anchors, or does it serve only to build rapport with the audience?
Rapport-building stories produce diminishing returns after the first ninety seconds. The audience already feels rapport with you because you are on a stage and they came to listen. Trying to build more rapport is like trying to make a guest more welcome after they have already sat down to dinner. The story-as-evidence is different. It anchors a claim with a real-world artifact the audience can verify or visualize. A story about a specific Microsoft team that ran a specific pilot in 2023, with the data, is evidence. A story about your grandfather that has charm but no claim attached is filler.
The best keynotes contain two to four stories total, each of which advances a specific anchor. The stories are stripped of decorative detail and engineered around the claim they support. The reader of a published essay can tolerate decorative detail. The listener of a live speech cannot, because the listener has no rewind button and limited attention bandwidth.
The close has to deliver the promise
The closing ninety seconds need to do three jobs. They need to restate the central argument in a form sharper than the opening. They need to make explicit the one action or change in thinking the audience should now take. They need to land a final sentence the audience will repeat. The “final sentence the audience will repeat” is non-negotiable. Without it, the room exits without a takeaway, and the speech evaporates regardless of how strong the body was.
The final sentence is engineered, not stumbled upon. It is short, declarative, and rhythmically distinctive. It often turns on a verb. It is usually written in the last fifteen percent of the writing process, after the rest of the speech is finalized, because the writer has to know what the speech is actually saying before the closing line can be shaped to land it. The closing line is also the most-rehearsed line in the speech. A speaker who fumbles the close has trained the audience to remember the fumble.
Rehearsal converts a written speech into a delivered one
A written keynote is not a delivered keynote. The conversion happens in rehearsal, and rehearsal is the step almost every first-time speaker shortcuts. Six to eight hours of full rehearsal, broken into discrete passes, is the floor. The first pass is for cuts. Read the speech aloud and cut anything that sounds wrong to your own ear. The second pass is for transitions, where one anchor moves to the next. The third pass is for pacing, where you mark the moments to slow down and the moments to speed up. The fourth pass is for the body language, especially the openings and closes. The fifth pass is recorded video, watched back at 1.5x speed, with notes on every visible verbal tic. The sixth pass is a live test in front of a small audience.
The single largest predictor of how well a keynote lands is rehearsal hours. Smart speakers rehearse the opening ninety seconds at least twenty times. They rehearse the closing ninety seconds another fifteen. They rehearse the full speech start-to-finish four to six times before the day. The result is not over-rehearsal that flattens spontaneity. The result is the freedom to be spontaneous because the structure is locked. A speech delivered cold can sound off. A speech delivered after thirty hours of writing and ten hours of rehearsal sounds inevitable. The audience cannot tell the difference between inevitable and effortless. That is the trick. The speech that looks effortless was the one that was hardest to write.