I have watched a lot of keynotes from the side of the stage, waiting to go on. You develop a fairly quick sense of which rooms are actually listening and which ones are being polite.
The difference is almost never the speaker's energy. Plenty of high-energy talks slide straight off an audience. It is whether the talk gave people a new way of looking at something they were already carrying around.
The going rate for a professional keynote in Canada. That buys you roughly forty-five minutes of the undivided attention of everyone you invited. It is the most concentrated attention your organisation will get all year.
The thing memorable keynotes have in common
They take something the audience already deals with and make it visible.
Not new information. Visibility. Most people in your room are not short of facts about their own industry; they are numb to them. A number they have seen on a dashboard four hundred times has stopped meaning anything. The job of the keynote is to make that number feel like it means something again.
I do this with carbon footprint, which is genuinely one of the least engaging datasets in existence. People know their emissions matter. They have known for years. Knowing has changed nothing, because the number is abstract and the abstraction is protective.
So I connect it to The Beatles, to Taylor Swift, to the coffee someone drank an hour ago. Not as decoration. As translation. Once carbon is measured in things you have opinions about, you start having opinions about carbon.
That mechanism has nothing to do with climate. It works on product metrics, churn, patient outcomes, learning results, safety incidents. Any dataset that people have gone blind to.
Four things that predict whether it lands
1. One idea, not seven
A memorable talk has a single spine. Everything else exists to make that spine unforgettable. Talks with five insights produce audiences with zero. Ask any candidate what the one idea is. If they cannot say it in a sentence, the room will not be able to either.
2. The audience does something
Not a gimmick. A moment where people have to commit to an answer, guess a number, or notice they were wrong. Being wrong in public, gently, is the fastest route to remembering. A room that has been surprised by its own answer is paying attention for the rest of the session.
3. It is built for this room, not a room
Ask what changes for your audience. The answer tells you almost everything. A speaker who says "I will swap in your industry's examples" is doing surface work. A speaker who asks which decision your people are avoiding is doing the real thing.
4. It ends with something checkable
Inspiration decays fast. What survives is a specific action small enough to do on Monday. If the closing minute is a rousing call to be better, the effect is gone by lunch on Tuesday.
What does not predict it
- Fame. Recognition fills a room. It does not hold one. Some of the flattest sessions I have watched were delivered by people the audience had queued to see.
- Slide production values. Beautiful decks are pleasant and forgettable. Nobody has ever left a conference talking about a transition.
- Sector pedigree. Useful for the first two minutes of credibility. Almost irrelevant afterwards.
- Length. Longer is not deeper. It is mostly just longer.
How to test for it before you book
Watch a full session, not a sizzle reel. Reels are edited to show reactions; they tell you the speaker can be filmed well. A complete talk tells you whether the argument holds for forty-five minutes, which is the actual purchase. Mine is on the speaker reel page and I would rather you watched the long version.
Then ask the question that separates the two kinds of speaker: what would you need to know about our audience to make this work? The good answer is a list of questions. The bad answer is reassurance.
There is a fuller list in the questions to ask before you book.
The honest caveat
Some events do not need a memorable keynote. Sometimes the room needs energy after lunch, or a name on the programme that helps sell tickets, and that is a legitimate thing to buy. The mistake is buying that and expecting behaviour to change afterwards.
Decide which one you need. Then book for it.
Frequently asked questions
How long should a keynote be?
Forty-five minutes including Q&A suits most conference programmes. Sixty works when the audience is seated and fed. Beyond that you are relying on the speaker's stamina and the room's air conditioning, and one of them will lose.
Are interactive keynotes better than straight talks?
For retention, usually yes, because participation forces attention. But interaction bolted onto a talk that has nothing to say is worse than a good lecture. The interaction has to carry the idea, not decorate it.
Does the speaker need industry experience in our sector?
Less than people think. Deep sector knowledge helps with credibility in the first two minutes. After that, what matters is whether they can make your specific numbers feel consequential. Outsiders are often better at that, because they can still see what insiders have stopped noticing.
How do you know afterwards whether it worked?
Ask a question with a checkable answer. Not 'did you enjoy it' but 'what will you do differently next week'. If most of the room cannot answer, the session was entertainment. That is sometimes what you wanted, but you should know which one you bought.
Sources: ProSpeakers - Canadian keynote fee ranges 2026 · Executive Speakers Bureau - what meeting planners should ask