Answer a safety question in three moves. Give a clear top line first, in one sentence the audience could repeat to a neighbor. Then give the nuance that makes the top line honest. Then say what the person should actually do. Opening with qualifiers reads as evasion, no matter how accurate the qualifiers are.
In a two-day session with a federal agency in Washington DC on 13 and 14 August 2026, a scientist in the room said the thing that most technical experts feel and almost never say out loud. Safety questions are hard, she said, because the honest answer is nuanced and the audience is asking for a direct one.
She was right, and the room agreed instantly. Everyone there could handle a hostile question. What they could not handle was a fair one they did not want to answer in a single word.
Why is a safety question so hard for a technical person?
Because the expert and the audience are asking two different questions using the same words.
When a member of the public asks "is it safe," they are asking a personal question. Should I let my kids do this. Should I change anything this week. Do I need to worry. It is a question about behavior, and it has a short answer.
When a scientist hears "is it safe," they hear a technical question. Safe at what exposure, over what period, measured by which method, against a standard that is itself under review. That question genuinely does not have a one-word answer, and the expert is not being evasive when they say so. They are being accurate.
The trouble is that accuracy delivered in the wrong order sounds exactly like evasion. So the expert reaches for "it depends," which is true, and the audience hears a person who is deciding whether to tell them something rather than a person who is telling them something. That is a delivery failure, not a science failure, and it is fixable in about ten minutes.
What does the qualifier-first answer actually sound like?
Here is a version I hear constantly from smart, careful people. Imagine a contamination story in a small town, and a reporter asks whether the water is safe to drink.
"Well, it really depends on the level of exposure and the duration, and I should say the sampling we have so far covers part of the system rather than all of it, and the standard itself is currently being re-examined, so I would not want to say one way or the other until we have more complete information."
Nothing in that answer is false. Every clause is defensible. And it is a disaster, because the first eleven words tell the audience that you are not going to answer, and after that they stop listening for information and start listening for motive.
Now the same facts, reordered. Nothing added, nothing dropped.
"Based on everything we have measured so far, we have not found a level that would change what I would tell my own family to do. Here is the part I want you to hear as well: our sampling covers part of the system, not all of it, and the standard we measure against is being re-examined right now. So here is what I would do this week. Use the filter if you have one, and check the posted results on Friday, because we will have more of the system covered by then."
Same information. Same caveats. Completely different reception, because the audience now has somewhere to put their worry before they have to hold the complexity. If you want the deeper version of this reordering habit, it sits next to the work in how to answer tough media interview questions.
What is the top line, nuance, next step construction?
Three parts, always in this order.
One: the top line. One sentence, no subordinate clauses, that a person could repeat accurately to a neighbor over the fence. It must be something you would defend under oath. It is not a simplification of the science, it is a statement of what the science currently supports. If you cannot write it, you do not yet know what you think, and that is worth discovering before a camera is pointed at you rather than during.
Two: the nuance. Now you get to be a scientist. Say what the top line rests on, what it does not cover, and what would change it. Signal it explicitly with a phrase like "here is the part I want you to hear as well," because that phrase tells the audience the qualification is being offered to them rather than hidden from them. Same content, opposite meaning.
Three: the next step. The single most skipped part, and the one that converts an answer into leadership. What should the person do between now and the next update, and when is the next update. Uncertainty is bearable when it comes with a date. It is unbearable when it comes on its own.
Practise it out loud in that order until it stops feeling backwards. It will feel backwards at first, because technical training teaches you to build to a conclusion and this construction leads with one. That reversal is the entire skill. The related work on picking the sentence you want repeated is in key messages for media interviews.
What does the strongest version of this sound like?
The best example I have found did not come from a training room. It came from a press conference recording, where an official was pushed hard for a yes or a no about whether a community was in danger. He refused both, and then he made the refusal worth something. He committed on the spot to a complete survey, every location, so that everyone could see exactly where they stood.
Notice what that line does. It does not scold the reporter for wanting a simple answer, it agrees that wanting one is reasonable. And it reframes the refusal as an upgrade rather than a withholding.
That last part is load-bearing. Refusal plus commitment reads as leadership. Refusal on its own reads as evasion. If you are going to decline the easy answer, you have to hand over something bigger in the same breath.
How do you find a top line when the data is genuinely mixed?
Ask yourself what you would tell a friend who called you tonight. Not what you would publish. What you would say on the phone. Then write that down and check it against the evidence. In twenty years of doing this I have almost never seen an expert who did not have a defensible top line, and I have very often seen one who had never been asked to say it in a single sentence.
If the answer really is that you do not know yet, then that is your top line. "We do not know yet" is a clear statement, and it is a great deal stronger than "it depends," because it is a fact about the situation rather than a hedge about your willingness to speak. Follow it with what you are doing to find out and when you will say more.
Does this only apply to a technical public-health risk?
No, and this is where I would push the point hardest, because every regulated industry runs the identical pattern with different nouns.
- Healthcare. "Is this treatment safe?" The honest answer involves populations, base rates and individual risk factors. The patient is asking whether to take the pill on Tuesday.
- Financial services. "Is my money safe?" The honest answer involves capital ratios, insurance limits and counterparty exposure. The customer is asking whether to move it in the morning.
- Food and beverage. "Is this recall going to affect what is in my cupboard?" The honest answer involves lot codes and distribution windows. The shopper is standing in front of an open cupboard.
- Insurance, energy, technology. "Is my data safe," "is the grid going to hold." Same structure every time.
In each case the expert's instinct is to explain the denominator before giving the number. Reverse it. Give the number, then the denominator, then the action. I have watched that single reordering change the tone of a story before the reporter had finished writing it down. It is also the reason so much of what agencies and companies call a message problem turns out on inspection to be a sequencing problem, which is where the work in crisis communication usually starts.
What did the room not settle?
I will not pretend this was resolved cleanly. Several scientists in that session pushed back on the top line requirement, and their objection was serious: a confident opening sentence can imply more certainty than the underlying evidence supports, particularly early on when you have a single round of sampling and nothing to compare it to.
My answer is that the top line does not have to be reassuring, it only has to be clear, and "we have one round of results and they are not enough to draw a conclusion from" is a perfectly good top line. But that exchange did not end with everyone nodding, and it should not have. If you work in a field where a clear first sentence can be read as a promise, argue this one out inside your own organization before the day you need it.
What should you take away?
Safety questions are not a test of your science. They are a test of your sequencing. Lead with a sentence someone can repeat, follow it with the nuance that makes it true, and finish with the thing they should do next.
If your experts field this kind of question and the coverage keeps coming out sideways, tell us about it here. One session on sequencing usually changes the next set of headlines.
Drawn from a two-day media training session with a federal agency in Washington DC, 13 and 14 August 2026. Participants and the subject matter of that session are not identified. The press conference quotation comes from a publicly available recording and is reproduced as spoken.