A chief executive I worked with had to record a four minute video telling roughly forty thousand people that a restructure was happening. He recorded it eleven times. By take eight he was word perfect, warm, well lit, and completely unconvincing.
The problem was not his delivery. The problem was that he did not believe the third paragraph, and forty thousand people were about to watch him not believe it.
That is the whole subject, and almost nobody teaches it that way.
What you have probably been taught
Somewhere in your career a presentation coach told you that seven percent of communication is words, thirty eight percent is tone and fifty five percent is body language. It is the most repeated statistic in corporate communication training.
It is also a misreading of two small studies from the late 1960s that measured something else entirely. Albert Mehrabian was testing how people judge whether someone likes them when the signals conflict: a single word, thanks, delivered warmly or coldly, with no sentence around it and no conversation taking place. The samples were small and entirely female. He was not measuring how information transfers, and he has spent decades saying so.
This matters because of what gets built on top of it. If ninety three percent of your message is delivery, then the way to improve is to perform harder: more animation, more gesture, more warmth than you feel. That advice is fine for someone talking to strangers on the internet. Applied by a chief executive to their own workforce it is close to the worst thing you can do, because your audience is not a stranger. They have sat in meetings with you. They know exactly what you look like when you mean it.
Performed warmth from someone the audience already knows does not read as warmth. It reads as management.
The other thing you were probably taught is to stand in a posture of expansion for two minutes beforehand. The lead author of that original study published a statement saying she no longer believes the effect is real, after a much larger replication found no hormonal or behavioural change. It is still in the deck of most executive communication programmes.
Two things that do hold up
You look far calmer than you feel. There is a well documented gap between how visible people believe their nerves are and how visible they actually are. Speakers consistently overestimate it by roughly half. The useful part is not the finding. It is that simply being told about it, before speaking, produced better speeches and less visible anxiety than being told to relax.
So the intervention is a sentence you say to yourself, and it is a fact rather than an affirmation: they cannot see it. Your heart rate is not on the outside. The thing you are certain everyone noticed at minute two was noticed by nobody.
Do not try to calm down. The instinct before a difficult address is to get the arousal down, and it does not work, because you cannot talk a nervous system out of being switched on in ninety seconds. Alison Wood Brooks tested the alternative: relabel the same physical state as excitement, out loud, in three words. Compared with people told to calm themselves, those who did were rated more persuasive, more competent and more confident.
The mechanism is worth understanding rather than just doing. Anxiety and excitement are the same arousal with different stories attached. One says threat, and threat makes people small, fast and defensive. The other says opportunity. You are not changing your body. You are changing what your body is evidence of.
What actually determines whether it lands
Now the part the research does not cover, which is what I have watched happen in rooms.
Say the difficult thing in the first thirty seconds. If there is news, everyone watching is scanning for it, and until it arrives they are not listening to anything else. Executives almost always bury it: context, then rationale, then the news at minute three. From the other side that structure does not read as considerate. It reads as someone building a case before an admission, which makes the news worse when it lands.
Talk to one person. Not to forty thousand, and not to the lens. Pick somebody real, ideally somebody two or three levels down whose face you can actually picture, and talk to them. The register changes immediately and everyone can hear it. This is the single fastest change available and it costs nothing.
Stop performing certainty you do not have. This is the one that matters most and the one most senior people get wrong, because they have been told that confidence reassures. What reassures is congruence. If you are not sure the restructure will work, say that you are not sure, and then say what you are sure of. An audience that catches you performing certainty stops believing the parts that were true, and they catch it more reliably than you think, because incongruence is the one thing humans are genuinely good at detecting.
Let the pauses stay. Silence feels three times longer to the person producing it than to anyone watching. The instinct is to fill it, and filling it is what produces the verbal padding that makes senior people sound evasive. A pause after a hard sentence is not dead air. It is the only signal available that you know the sentence was hard.
The rehearsal trap
My client's eleven takes were the problem, not the solution.
Rehearsing to the word produces recital, and recital is instantly detectable: the eyes go slightly wrong as the person reads ahead internally. Rehearse the structure instead. Four beats, in order, that you could deliver in any words. Then let the words be different every time, including on the take you keep.
Take twelve was the one we used. It had a stumble in it, he lost his place once, and he said the thing about not being certain. It is the only version anyone remembered.
The question underneath
There is a reason this subject fascinates people at your level, and it is not vanity. It is that somewhere between one hundred and a few thousand people, you stop being able to persuade anyone in person. The room becomes a screen. Everything you communicate arrives mediated, edited, forwarded, and watched at 1.5 times speed on a phone by someone deciding whether to update their CV.
At that scale, presence is not a soft skill sitting next to the real work. It is the mechanism by which the real work reaches anybody.
So the question is not how to look more confident on camera. It is a harder one: what are you currently saying to your organisation that you do not entirely believe? Because whatever it is, they have already noticed, and no amount of lighting is going to help.