Knowing how to answer questions you cannot answer starts with one move: name why the information is missing. A missing fact, an uncertain forecast, an expertise limit, and a privacy boundary each call for a different reply.
How to answer questions you cannot answer: where do you start?
What caused the blank shapes a good reply. If no one checked a number, you need a record and a later check. If a forecast is uncertain, you need assumptions and limits. If the question is outside your field, you need a qualified source. If it touches private details, you need a boundary.
Before you name the cause, check the question itself. A question can sound exact while hiding two requests. "Can we hit the Friday date?" may mean the internal build or the customer count. Ask which one the person means. That one check can stop you from solving the wrong problem.
Research on unknowable questions suggests the cause matters. In three experiments with children and adults, older children and adults often preferred an informant who gave a reason precision was impossible. They favored confident answers for knowable facts instead (Kominsky et al., 2016). Those were judgments about informants, not audience behavior.
Does admitting uncertainty always reduce trust?
No. But the evidence is narrow.
Five experiments tested this. They included a preregistered national sample and a BBC field experiment. Trust dropped only a little after uncertainty. The drop mainly involved verbal uncertainty. Numeric uncertainty often cost little trust (van der Bles et al., 2020). These studies looked at written statistical messages. They did not test a live answer during a talk. They do not promise that uncertainty keeps trust in every room.
Credibility also depends on fit, not confidence alone. Two controlled experiments used conflicting written eyewitness accounts. Once accuracy evidence appeared, judgments tracked whether stated confidence matched actual accuracy. They did not track confidence by itself (Tenney et al., 2008). The design used simulated juror judgments, not business Q&A.
Together, these findings point away from sounding sure. They point toward a proposed standard: sound calibrated. Calibrated means your certainty matches the evidence you have.
What can you say when a fact needs checking?
Separate what you know from what you are about to find out. When a question needs a verifiable fact, the useful reply has three parts: the honest status, the source you will check, and when and where you will report back.
Suppose someone asks whether every customer will move to a new platform by Friday. This is a hypothetical illustration, not a client story. Internally, the platform passed its tests. What has not been verified is the customer migration count. Those are two different facts; one does not establish the other.
A workable reply might sound like this: "The platform passed its internal tests. I do not have the verified migration count in front of me. The operations lead owns that number. I will share a status update at Thursday's session."
Commit only to a step you can take. In the example above, the speaker must be able to ask the operations lead for a status. Confirm that before you speak. If the owner is unavailable or the data are restricted, say what you can still provide.
How should you handle forecasts, expertise limits and privacy?
These three causes share one rule. Never let the shape of the answer pretend to be certainty. A forecast, an outside field, and a boundary each need language that matches what you truly have.
State the assumption the number rests on. Then state the limit. "If adoption holds at the current rate, we reach the target in the third quarter. That depends on two supplier timelines I cannot control."
An expertise limit is about who can properly respond. You might say you will ask a qualified colleague. You do not state that they can or will. A confidentiality limit means you cannot discuss individual cases. If allowed, you may share only authorized public process information. Giving no further detail can be appropriate.
In two preregistered experiments with UK adults on vaccination and nuclear power, balanced messages were as trusted as persuasive ones on vaccination. They were more trusted on nuclear power. Prior beliefs mattered, and trust did not always move with support for the proposal (Kerr et al., 2022). Several transparency features changed together, so disclosure alone was not isolated. Honesty is not a lever you can pull for a guaranteed reaction.
The table below is a proposed planning aid, not a validated protocol. Use it to sort a question before you answer.
| Cause | Useful reply | What to avoid |
|---|---|---|
| Missing verifiable fact | Name the record, the owner, and a follow-up time | Guessing the number |
| Uncertain forecast | State assumptions and limits | Stating the forecast as fact |
| Outside expertise | Name a suitable expert or source | Improvising outside your field |
| Confidential matter | Explain the boundary; give safe general information only if authorized | Sharing restricted information |
How can you make a follow-up promise useful?
A promise to check is not a promise that an answer exists. That distinction keeps you honest.
Keep a small log after any session with open questions. One optional format uses seven columns. These are question, factual boundary, owner, safe channel, due date, status, and correction note. The dates you choose are examples. They are not research-based intervals. Pick what your work allows. Review the log on a set day. Do this until each line closes or is formally set aside.
Keep the public log free of sensitive employee details. A shared document is not the place for names in a personnel matter. Route those separately and report only the outcome that is safe to share.
When the channel matters, say it. "I will post an update in the team channel on Tuesday." If the answer may not exist, say that too. "If the figure cannot be verified by Tuesday, I will still give you a status on Tuesday." This commitment remains meaningful even when the data do not arrive.
What should you do if your first answer was wrong?
Find the earlier error. Then send the verified correction or revised uncertainty through an approved channel. Reach affected people where you can. If you said 40 migrations were done, and checked records show 30 at the same cutoff date, state that correction. A count alone does not promise a future completion date. An estimate is not a forecast on its own. It needs a basis you can defend, stated uncertainty, and fit for the decision. Checks may not be available.
For event fit, see speaking enquiries. The Insights library covers leadership training transfer and interpreting 360 feedback.
Key takeaways
- First, find the cause. It may be a missing fact, an unsure forecast, outside expertise, or a privacy limit. Each cause needs a different reply.
- Match how certain you sound to the evidence you have. Give the reason the uncertainty exists.
- If a fact is missing, name the record, the owner, and the channel.
- A promise to check is not a promise an answer exists. Say so.
- Correct errors through an appropriate authorized channel that reaches affected people where possible.
Frequently asked questions
References
- Kerr, J. R., Schneider, C. R., Freeman, A. L. J., Marteau, T., & van der Linden, S. (2022). Transparent communication of evidence does not undermine public trust in evidence. PNAS Nexus, 1(5), pgac280.
- Kominsky, J. F., Langthorne, P., & Keil, F. C. (2016). The better part of not knowing: Virtuous ignorance. Developmental Psychology, 52(1), 31-45.
- Tenney, E. R., Spellman, B. A., & MacCoun, R. J. (2008). The benefits of knowing what you know (and what you don't): How calibration affects credibility. Journal of Experimental Social Psychology, 44(5), 1368-1375.
- van der Bles, A. M., van der Linden, S., Freeman, A. L. J., & Spiegelhalter, D. J. (2020). The effects of communicating uncertainty on public trust in facts and numbers. Proceedings of the National Academy of Sciences, 117(14), 7672-7683.
This article is for informational and educational purposes only and does not constitute financial, legal, tax, medical, or professional advice. Individual results vary.