How to Know a Conversation Worked

Published October 23, 2026 · Last reviewed October 23, 2026

Why your own feeling is the wrong instrument

Keysar and Henly (2002, Psychological Science) asked speakers to say ambiguous sentences while trying to convey a specific meaning to a listener, then to estimate how often the listener understood. Speakers showed a consistent tendency to overestimate: they expected to be understood more often than they were. Overhearers who were told the speaker’s intention did not systematically overestimate. The authors conclude that speakers underestimate the ambiguity of their own words, because they already know what they meant.

Gilovich, Savitsky and Medvec (1998, Journal of Personality and Social Psychology) found a related pattern across three sets of studies: people tend to overestimate how well others can read their inner states. In the abstract’s examples, liars overrated how likely their lies were to be caught, and people overrated how visible their disgust was.

Limits. Both are laboratory studies. Keysar and Henly used ambiguous sentences, so how large the effect is in your real conversations is Not reported. The safe takeaway is a direction: you probably feel clearer than you were, so look for evidence.

Signs you can observe, and signs you cannot

A nod and silence belong in the right-hand column. Both fit understanding, confusion and politeness equally well.

The 24-hour check

1. Decide the sign in advance. Before the conversation, write the one behavior that would count: “She restates the new deadline correctly,” or “He tells me which of the two options he picks.”
2. Ask for their version before you part. Say: “I want to check I was clear. How would you describe the plan to someone who was not here?”
3. Put the decision in writing within the day. Send two lines that state what was agreed, and invite correction: “Tell me where I have it wrong.”
4. Look at what happens next. By the next day, did they act, reply or ask a question that shows they used what you said?

Say it out loud

“I tend to think I am clearer than I am. Could you tell me back what you are taking from this, so I can fix anything I left out?”

The request puts the check on your clarity, not on their attention, and you can state it openly. A listener who knows why you are asking loses nothing by answering.

Profile translations

• Profile A After a conversation about chores, money or plans, ask the other person to say what happens next in their words. Compare their version with yours.
• Profile B End each one-on-one with “what are you doing first?” and note whether the first action matches what you meant.
• Profile C After teaching, ask one learner to explain the idea to a neighbor. The first confused question is your data.
• Profile D After a sales call, send a three-line recap and see whether the buyer corrects it. A correction is a sign the buyer read it.
• Profile E After a long exchange online or at a meeting, ask a participant to name the other side’s strongest point, and see whether they can.

The practice pages go deeper: Listening before deciding covers paraphrase in a one-on-one, and Conversational receptiveness covers disagreement. The conversation planner lets you write the sign you will watch for before you start.

A worked example

An illustration, not data. A manager ends a meeting believing the new review schedule is clear. The two ways to find out look like this:

The second question asks for a fact that can be right or wrong, so a misunderstanding has somewhere to show up. Questions that invite only “yes” give it nowhere to go.

When you cannot ask in person

In email and chat, the same logic applies in writing. End the message with one concrete question that can only be answered by someone who read it: “Which of the two options do you prefer, and by when?” A reply that answers it is a sign of reading. A bare “thanks” carries no information about understanding, so wait for the next observable sign before you treat the matter as closed.

How to read the result

An inaccurate paraphrase is useful information, not a failure. It shows exactly which part of your message travelled badly. Restate that part in different words and ask again. If nothing observable happens within a day (no question, no action, no reply), treat the conversation as unfinished.

Failure signs: you hear only “sounds good” with no detail, the same question returns in a week, or the other person acts on a different version of the plan. Each one means the shared picture was never built.