Where we are
Session 1 gave you a card: one sentence, three supports, a cut list and a word budget. Noor Haddad, our constructed team lead, has hers: "From Monday we answer the customer who has waited longest first", with the six-day wait, the average that hid it, and what changes on Monday hanging from it. Tonight she writes the two ends of her ten-minute all-hands talk, the parts a room remembers best and speakers prepare least. Noor, her numbers and her lines are invented for teaching. You write yours alongside.
★ What you walk out with tonight. A written opening of about 45 seconds that answers the four questions a room has before it decides to listen, with every thank-you, agenda and apology struck out; a written close that returns to your sentence, says your ask and stops; the bridge line that puts questions before that close; and both ends said out loud and timed.
Give the room a reason to listen
6 min liveA meeting has a reason to listen built in: people came because the topic is theirs. An all-hands does not. The room is captive, half of it is on a laptop, and nobody chose your ten minutes. Anderson's second guideline is written for exactly this: before you build anything, give the listeners a reason to care, by sparking their curiosity about the thing you are about to explain.
LiveFive ways to spend thirty seconds on nothing3 min▶
Each of these feels polite or safe from the stage. From the seats, each one delays the only thing the room wants to know, which is why it should keep listening.
- Thanking the organiser. The room already knows who invited you. Thank people at the end of the event, in person, or in the follow-up note.
- The agenda slide. "First some background, then the data, then next steps" tells the room the order of your work, not the reason for it. A ten-minute talk does not need a map; three supports are the map.
- The apology. "I'm not a natural speaker, so bear with me" asks the room to watch for exactly the thing you are worried about. It also spends goodwill you have not used yet.
- The context history. "Quick bit of context: back in March we started looking at..." is the X item from your cut list, coming back in through the front door.
- The borrowed joke. A joke that is not about your subject buys a laugh and then a reset, because the room now has to find out what the talk is actually about. If something genuinely funny comes out of your material, keep it.
None of these is a research finding; they are the openings that waste the slot in practice, and the test is simple: does the sentence answer one of the room's four questions?
LiveThree openings that do work, each built from Noor's card3 min▶
All three are built from material already on the card. The opening is not new content; it is the most interesting part of what you have, moved to the front.
- The specific moment. "One week in May, our average reply time was under four hours. That same week, a customer in Lisbon waited six days for a refund reply." One time, one place, one person, one number. Curiosity comes from the gap between the two facts.
- The question the room already has. "If our average reply is under four hours, how does a customer wait six days?" It works only if the room would genuinely ask it once they heard the two numbers; a question nobody has is a quiz.
- The claim, said plainly. "From Monday, the customer who has waited longest gets answered first." The throughline itself, first. It is the bravest of the three and the right one when the room is senior and impatient.
The four questions in the figure (who is this, why care, how long, what do I do with it) are a working model for checking an opening, not a measured fact about audiences. The meeting-length version of an opening, the ask plus the shape and the pause after the number, is taught in Communication, session 5 ↗. This page does not repeat it: a room-sized talk needs the same discipline plus a reason to care, because the room did not come for you.
Self-studyNoor's rewritten opening, word by word2 min▶
Constructed, 58 words: "I'm Noor, I run support operations. One week in May, our average reply time was under four hours. That same week, a customer in Lisbon waited six days for a refund reply. From Monday, we answer the customer who has waited longest first. Ten minutes: why the average hid her, what changes, and one thing to watch for."
Who she is takes six words. Why the room should care takes the two facts and the sentence. How long is two words. What to do with it is a promise of one thing to watch for, which the close will pay off. At her rehearsed pace of 126.6 words a minute that is about 28 seconds, well inside the 45-second opening her session 1 budget allowed.
Anderson opens his own talk with the claim, not a welcome. The opening line of the published transcript of "TED's secret to great public speaking" is "Your number one task as a speaker is to transfer into your listeners' minds an extraordinary gift". No thanks, no agenda: the room learns in one sentence what the talk will argue. Ken Robinson does the same thing early in his 2006 talk with "My contention is that creativity now is as important in education as literacy", and every story after it hangs from that line.
The last line is the one they carry out
5 min liveMost talks do not end. They stop, usually on "so, yeah, that's all I have, any questions?", and then the last thing the room hears is the last question, which may be about something you cut. The fix is a matter of order: take the questions before your final line, then return to your sentence, say the ask, and stop.
LiveWhat the close has to do3 min▶
Three jobs, in this order, and nothing after them:
- Return to the throughline. Say the sentence from your card again, in the same words. Repetition that feels clumsy on the page is what a listener needs: it tells them which of the last ten minutes was the point.
- Say the ask. One thing the room should do, think or watch for. Noor's (constructed): "When the average rises next week, ask me about the two-day number instead." It turns a coming bad number into something the room expects, and it pays off the "one thing to watch for" her opening promised.
- Stop. No "so, yeah". No "that's all I have". No summary of the summary. Stand still for a beat after the last word; the room will know it was the end because you did.
LiveQuestions go before the last line2 min▶
Questions are not the end of a talk; they are a section of it. Announce them after your last support: "Before I finish, any questions?" Take them, and when time is up, call the last one yourself. Then bridge back: "Let me leave you with the one thing I'd like you to remember." Then the sentence, the ask, and stop.
This order costs nothing and protects the most valuable seconds of the slot. Without it, the last thing a room hears is decided by whoever had the last question, and it is often about the detail you cut on purpose. Handling hard questions well is session 6; tonight is only about where they sit.
Self-studyWhen there is no time for questions at all2 min▶
All-hands slots often run late and the host cuts questions. Write the close so it stands without them: your last support leads straight into "So:" and the sentence. Then point to where questions can go instead ("my door, or reply to the note I'll send this afternoon"), and that line goes before the ask, not after it. The note is the one-page handout you built from your cut list in session 1.
A familiar town-hall ending. The last slide says "Questions?", the final question is about the car park, the host says "great, let's leave it there", and a meeting about next year's strategy ends on parking. Nothing went wrong in the talk itself. The order did it. Speakers who put questions before their final slide, and then return to one sentence, leave the room with the strategy instead.
Words are not 7% of your talk
4 min liveSomewhere in your career someone will tell you that only 7% of communication is words, 38% is tone and 55% is body language, usually to explain why your opening should be all energy and eye contact. The numbers are real. The rule built on them is not, and the person who produced the numbers says so himself.
LiveWhat Mehrabian says the numbers mean2 min▶
Albert Mehrabian's own summary page for his book Silent Messages (1971; 2nd edition 1981) gives the formula as "Total Liking = 7% Verbal Liking + 38% Vocal Liking + 55% Facial Liking". It is about liking: how much a listener concludes the speaker likes them, or likes what they are talking about.
He says it was derived "from experiments dealing with communications of feelings and attitudes (i.e., like-dislike)", and then draws the boundary himself: "Unless a communicator is talking about their feelings or attitudes, these equations are not applicable."
An all-hands talk explaining a change to how tickets are answered is not a communication of like or dislike. By the author's own statement, the formula has nothing to say about it.
LiveWhat follows for a speaker, honestly2 min▶
Two things are true at once. Tone and face matter a great deal for whether a room trusts you: a flat voice saying "I'm excited about this change" will not be believed, and that is roughly the situation the formula describes, words and tone disagreeing about a feeling. But the content of your talk is not 7% of anything. If Noor's words are wrong, no amount of eye contact makes "oldest first" make sense to the room.
So the opening and close you write tonight still matter more than how you stand while saying them. Voice and body technique are not taught on this page, because no read-at-source evidence for specific rules was assembled; session 3 deals with the nerves that make delivery wobble.
Self-studyHow a narrow finding becomes a universal rule2 min▶
The original work is two papers from 1967 (Mehrabian and Wiener; Mehrabian and Ferris). Neither could be opened for this page, so nothing here describes what their participants heard or saw; the claims rest only on Mehrabian's own later summary, quoted above.
The pattern is worth knowing because it repeats. A finding is published with its scope: who was studied, what they judged, under what conditions. It gets summarised as three numbers. The summary gets repeated without the scope, then repeated again as "93% of communication is non-verbal", which is a sentence the author's own limit rules out. At each step it gets shorter, more confident and more general, and nobody along the chain is lying; each one just drops the part that made it true.
The defence is a habit, not a database: when someone quotes a number about speaking, ask what was measured, on whom, and what the author says it does not cover. Session 3 runs the same check on "people fear public speaking more than death".
The researcher behind the number corrects the number. Mehrabian did not leave the limit for others to discover: it sits on his own page, next to the formula. The 7-38-55 rule people repeat is not a misreading of a hidden footnote; it is the formula with its own limit sentence removed. That is the version to correct, politely, the next time it appears in a training deck.
Write both ends of your talk
★ 22 min · everyone buildsSix steps on your own talk, using the card from session 1, with Noor's constructed versions alongside. Write every line out in full. Openings and closes are the two places where speakers improvise most and should improvise least: they are short enough to know word for word, and they are what the room remembers.
Answer the room's four questions for your talk. One line each. If you cannot answer "why should I care" in a line, that is the opening problem, and the next step will not fix it until you can.
Write three candidate openings, then pick one. One of each kind: the specific moment, the question the room already has, the claim said plainly. Read them aloud. Pick the one you would want to hear if you were in the third row with a laptop open.
Strike every word of thanks, agenda and apology. Write your first instinctive opening, then cross out anything that does not answer one of the four questions. What is left is usually a single sentence. Build the chosen candidate onto it.
Write the close: return line, ask, final line. The return line is your card's sentence, word for word. The ask is one thing the room does, thinks or watches for. The final line is the ask itself or one short sentence after it, and then nothing.
Place the questions before the final line, and write the bridge. Mark the point after your last support where you open questions, and write the sentence that brings the room back to you when they end. You need it written, because you will say it right after a question that may have rattled you.
Say both ends out loud and time them. Standing, at your real speaking volume, with a timer. Your session 1 budget gave each end about 45 seconds, roughly 114 words at 152 words a minute; if you know your own pace, use it. Over budget means a cut, not speaking faster.
Add both ends to the card. Write the opening at the top and the close at the bottom, around the three supports, so the card now holds the whole talk's shape. Session 4 measures your recorded pace, and session 6 rehearses these two ends more than anything in between.
Before session 3
30 min- Say your opening to one person who does not know the topic, then ask them which of the four questions they could answer. Rewrite whatever they could not.
- Learn both ends word for word. Not the middle; just the first and last 45 seconds. Say them on a walk until you can start either one without looking.
- Watch the last minute of three recorded talks or meetings. For each, write down the very last sentence said. Count how many end on the speaker's own point and how many end on a question or "that's it".
- Find one place 7-38-55 is quoted as a rule (a training slide, an article, a post). Check whether the quote keeps Mehrabian's limit. Write one polite sentence you could say to correct it.
Three questions
Clear all three and session 2 gets stamped in your passport.
1 · Which opening gives an all-hands room the best reason to keep listening?
C is a specific moment that creates a gap the room wants closed, and it answers "why should I care" in two sentences. A and B spend the slot on nothing the room asked about; D asks the room to watch for the thing you are worried about.
2 · Where do questions go relative to your final line, and why?
Questions are a section of the talk, not its ending. Take them after the last support, call the last one yourself, then bridge back to the sentence, say the ask and stop. B and D let the last question decide what the room carries out.
3 · By Mehrabian's own statement, what does the 7-38-55 formula apply to?
His own summary page says the equations come from experiments on communicating feelings and attitudes, and that unless a communicator is talking about those, they do not apply. A and B are the formula with that limit removed.
What this session teaches, and where it came from
Every fact on this page, and its verification tier, is recorded in the course's source map.
Session 2 cheat sheet · pin this
You now have both ends written and timed. Session 3 deals with what happens to your body when you stand up to say them: why a racing heart is arousal rather than a verdict, what the "say I am excited" study found, what happened when someone tried to replicate it, and where the famous "people fear speaking more than death" ranking actually came from. Bring your card; you will say the opening again.