You have a rambler on your team. What actually works
10 September 2026
Standing agendas, time limits and asking someone to write it down all manage the symptom without changing the behaviour, because the person cannot hear themselves do it. What changes it is a specific named habit rather than a general request, feedback close to the moment, and some way for them to see their own words afterwards.
An engineering manager described it in a thread that ran to fifty-odd comments:
I’m an EM, and my team has a rambling problem. Of my 8 engineers, 3 of them are smart but incapable of communicating clearly or concisely. They ramble away in meetings, then when I bring it up in 1:1’s they apologetically agree. Then rinse wash repeat.
Across twenty-six replies, every suggestion was a process hack. Standup templates. Strict time limits. A parking lot. Ask them to write it in an email instead. One manager brought a kettlebell into the room and passed it to whoever was speaking.
Nobody suggested anything that changes the behaviour. All of it manages the symptom inside the meeting, which is worth doing, and none of it survives the meeting ending.
Why the 1:1 keeps failing
Notice the shape of the failure. You raise it, they agree sincerely, nothing changes, you raise it again.
That is not resistance. If it were resistance you would see defensiveness. What you see is apologetic agreement, which means the feedback is landing and still not converting.
The reason is that they cannot hear themselves do it. From inside a meeting, the caveats feel necessary, the context feels required, and the time spent talking feels much shorter than it was. They accept your description of a meeting they remember completely differently, and then walk back into the next one with no way to tell it is happening again.
Repeating the conversation adds guilt without adding capability.
What actually moves it
Name one behaviour, not the quality. “You ramble” is a description of them. “Give the answer in your first sentence, then the context” is a rule they can follow on Tuesday. The second one is small, testable and impersonal. Pick the single behaviour that costs the most and ignore the rest for a month.
The usual candidates: answering last instead of first, restating a point that already landed, and thinking out loud in a status update.
Close the gap between the behaviour and the feedback. A 1:1 on Thursday about a meeting on Monday asks someone to accept a description of an event they remember differently. In-the-moment is much more effective, and it does not have to be brutal. Agree a private signal, or a short message straight after. “That answer started at 90 seconds” the same afternoon does more than a paragraph a week later.
Give them a way to see it themselves. This is the step that is nearly always missing and the one that closes the loop. Their own words, after the fact, in a form they can look at. Written summaries after key meetings work well and cost nothing: writing forces the compression that speech does not, and the gap between what they said and what they meant becomes visible to them rather than asserted by you.
Make the standard explicit for everyone. If updates are supposed to be two minutes with the headline first, say so as a norm rather than as a correction of one person. It removes the singling out and it usually improves two or three other people quietly.
The line worth not crossing
Do not record someone to prove a point.
The moment this becomes surveillance you have traded a coaching relationship for a compliance one, and you will not get it back. Consent rules also vary by jurisdiction, and in several places all parties must agree.
If a recording is going to exist, it should be theirs. Their account, their meetings, their choice to look. The distinction between a mirror someone holds and a camera you point is the whole thing.
When it is not fixable
Be honest about the difference between someone who cannot yet and someone who will not.
Someone who accepts the feedback each time and still cannot change it usually lacks evidence, not motivation. Someone who disputes the premise every time is telling you something different, and the conversation is about expectations rather than skill.
Worth remembering while you are deciding: communication feedback blocks more promotions than technical feedback does. Saying nothing is not the kind option.
Questions
- Why does the conversation not stick?
- Because agreement is not the blocker. People generally accept the feedback sincerely and then cannot act on it, since the behaviour is invisible from inside the meeting where it happens. Repeating the conversation adds guilt, not capability.
- Is it fair to raise this at all?
- Yes, and withholding it is worse. Communication lines block promotions more often than technical ones, so a manager who says nothing lets someone hit a ceiling without knowing why. The obligation is to be specific, not to be silent.
- Should I record meetings to show them?
- Not unilaterally. Anything that looks like monitoring destroys the trust the conversation depends on, and consent rules vary by jurisdiction. If a recording is going to exist it should belong to the person being coached, on their own account, at their discretion.
- What if it does not improve?
- Distinguish unwillingness from inability. Someone who accepts the feedback each time and still cannot change it usually lacks evidence rather than motivation, and the answer is a tighter loop rather than a firmer conversation.
Read next
- How to stop rambling in meetings
Rambling is not a discipline problem. Speech has no revision loop, so the first draft is the only draft. Here is what to change given that.
- "You need to be more concise" — what they are actually reacting to
Being told to be more concise is rarely about total words. It is about how long the listener waited for the point. Here is what people are reacting to and what to change.
- "Improve your communication skills" on a performance review — what to do next
A review line is not a plan. Here is how to turn the vaguest sentence in performance management into something specific enough to work on and prove.