Journal

Structural Patterns

He guessed a quarter.

Six people in the room, ninety minutes, and he was sure he had barely spoken. The transcript said fifty eight percent of the words were his.

6 min read

A product lead recorded a planning meeting because he wanted to settle an argument about what had been agreed. Six people, ninety minutes. Before he opened the transcript he wrote down what he thought his share had been, mostly as a joke. He put twenty five percent, and he remembers thinking that was probably generous.

It was fifty eight.

The second number was worse. In ninety minutes he had asked three questions, and two of them were about the calendar.

Why the guess is always low

Your own speech is stored in far higher resolution than anybody else's. When you talk, you get intent, word choice, the small correction halfway through, the feeling of being heard or not. When somebody else talks, you keep roughly what they meant.

So when you reconstruct the meeting afterwards, you are counting detailed memories against thin ones, and it comes out wrong every time in the same direction. This is not a flaw in him. It is close to universal, and it is the reason self assessment is useless here.

Feeling like you listened is not evidence that you listened. It is evidence that you were not speaking, which is a different thing.

What a share should look like

There is no single right answer and anyone offering one is selling something. Airtime only means anything against the role you were playing.

  • A catch up between two people: roughly even, drifting either way depending on who has news.
  • A one to one where you are supporting somebody: comfortably under half. Fifty percent in someone else's bad week is a problem.
  • A meeting of six: an even share is about seventeen percent. People who feel they hardly spoke are often at thirty five or forty.
  • A briefing or a presentation: high share is the entire point. Nothing to fix.
  • A negotiation: usually the person talking less is doing better, because they are learning faster than they are revealing.

Two numbers that tell you more

How long your turns run

Forty percent of the airtime taken across twenty short turns is a conversation. The same forty percent taken in three long ones is a lecture with pauses in it. Long turns crowd people out even when the arithmetic looks fine, because somebody who has been waiting ninety seconds for a gap has usually abandoned whatever they were going to say by the time it arrives.

How many turns opened something

This is the best single proxy for whether you were in a conversation or delivering into one. It is also the number that surprises people most. A forty minute discussion containing one genuine question from you is common, and it does not feel that way from the inside at all.

When talking a lot is fine

Worth saying plainly, because the guilt around this topic is out of all proportion. Talking a lot is not a character defect. Somebody has to carry a conversation, silence is not automatically generous, and an interesting person is under no obligation to ration themselves down to an even split.

The thing to watch is whether your share moves. A share that changes with the room is a person responding to it. A share that sits at sixty five percent in every conversation regardless of who is there and what it is for is a default, and defaults are worth knowing about because you did not choose them.

You cannot count this live

Tracking your own share in real time takes enough attention that you stop following the conversation, which defeats the point and makes you a worse participant while you do it. Several people have told me they tried this and simply became quiet and distracted, which is not the same as being a good listener.

The only method that works is measuring afterwards. Record one ordinary conversation, separate the speakers, look at share of words, average turn length, and questions asked. Then compare it against the number you would have guessed beforehand. The gap is the actual finding, and it is usually more useful than the airtime figure itself.

Once is normally enough. Nobody needs to keep counting forever. They need to have counted once, because after that the estimate stops being a guess. The product lead has not recorded a meeting since. He does not need to. He knows roughly where he sits now, and knowing it turns out to be most of the fix.

Common questions

How much should I talk in a conversation?
There is no universal number, because the right share depends on what the conversation is for. In a two person catch up, roughly even is reasonable. In a one to one where you are supporting somebody, well under half. In a meeting of six an even share is about seventeen percent, and people who feel they barely spoke in such a meeting are often at two or three times that.
Why do I underestimate how much I talk?
Because speaking is encoded much more strongly than listening. Your own turns arrive with intent, word choice and self monitoring attached. Other people's turns are stored as gist. Reconstructing the conversation later, you weigh vivid memories against faint ones, and the ratio comes out wrong in a consistent direction.
Is talking a lot actually a problem?
Only when it is unintentional or one directional. High airtime is correct when you are the person with the information. It becomes a problem when it is your default across every conversation regardless of role, when it comes with frequent interruption, or when the quieter people in a group have stopped trying to get in.

Reading about it is the easy half.

The Listening Fox records a real conversation and shows you whether you actually do this, or only agree with it.

Coming to the App Store