How to close a training needs conversation with stakeholders: read back three lines, and ask which one is wrong.
Read this if you run the first training needs conversation with stakeholders. Skip it if you only ever receive finished briefs.
The short version. Before the call ends, read back three lines and ask which one is wrong. "Does that sound right?" gets a yes. "Which of these is wrong?" gets the truth.
4-minute read
In the last five minutes, say: "Let me read back three lines. Tell me which one is wrong."
If they correct a line, you've won. You found the misunderstanding while it still costs nothing to fix.
In 2002, psychologists Boaz Keysar and Anne Henly had speakers say ambiguous sentences with one meaning in mind. Then the speakers guessed whether their listeners had got it. Speakers expected to be understood more often than they were. [1]
Overhearers who were told the intended meaning didn't make that mistake. Only the person speaking did. [1]
Keysar later called this a systematic cause of miscommunication, not just noise. [2] A training request works the same way: a speaker with a meaning in mind. "Our managers need resilience training" can point at very different problems. You both leave the call sure you agree.
Here's what changed. More calls now end with an automatic summary in everyone's inbox. Zoom says monthly users of its AI Companion more than tripled in its 2026 financial year. [3]
A summary is built from what was said. It can't check what was understood. If the sponsor said "resilience workshop", the summary says "resilience workshop". And now that's the official record.
So the read-back matters more now, not less. It's the one part of the call a notetaker can't do for you.
Same call, two endings. (An illustrative composite, not a client.)
| Before | After |
|---|---|
| "Great, I'll scope a two-day resilience workshop for the team leads. Does that sound right?" | "Three lines. Tell me which is wrong. One: when priorities change mid-week, team leads go quiet instead of renegotiating deadlines. Two: a programme can build the renegotiation skill; you own whether deadlines are allowed to move. Three: we'll know it worked when the project office sees fewer missed handoffs by March." |
| Sponsor: "Yes, perfect." | Sponsor: "Two's wrong. Deadlines can move. Nobody believes me." |
That one correction changed the brief. Part of the fix is now a message from the sponsor, not a workshop. You found it in ten seconds, not ten weeks.
In my experience, asking which line is wrong makes disagreement the expected answer. A plain "yes" now has to be a real one.
If they say all three are right, ask once more: "If one of these turns out wrong in three months, which will it be?"
Watch line 2 closely. If the sponsor can't name a person for the part a programme can't fix, nobody owns it yet. That's worth knowing before you design anything. For redirects that help here, see six moments where training-request calls go wrong.
If line 3 stays blank, you don't have a measure yet. Build a Level 3 plan before you promise one.
Put your three lines at the top of the follow-up note. Let the automatic summary sit below them as the transcript.
One is the record of what was said. Yours is the record of what was agreed. Ask for one reply: "Which line would you change?"
Want to rehearse the conversation that comes before the read-back? The free Diagnostic Conversation Coach walks you through it. No signup, no email.
The Read-Back Close · copy this
Say: "Let me read back three lines. Tell me which one is wrong." 1 · The problem, as behaviour When ________, people ________ instead of ________. 2 · The split A programme can fix ________. ________ (a named person) owns ________. 3 · The sign and the date We'll know it worked when ________ sees ________ by ________. If they say all three are right: "If one turns out wrong in three months, which will it be?" Within the hour: send these three lines at the top of your follow-up, above any AI summary. Ask: "Which line would you change?"
[1] Keysar, B., & Henly, A. S. (2002). Speakers' overestimation of their effectiveness. Psychological Science, 13(3), 207–212. doi.org/10.1111/1467-9280.00439. Claims here are limited to the published abstract, read 20 October 2026.
[2] Keysar, B. (2007). Communication and miscommunication: The role of egocentric processes. Intercultural Pragmatics. doi.org/10.1515/IP.2007.004. Abstract read 20 October 2026.
[3] Zoom Communications, Inc. Fiscal 2026 Annual Report (fiscal year ended 31 January 2026), shareholder letter: "AI Companion MAUs more than tripled during FY26." sec.gov.
The before/after exchange is an illustrative composite, not a client conversation.