Wrivio
Get Wrivio
7 min readBy Wrivio Team

How to Say You Are Not Sure in a Work Message Without Sounding Unprepared

Someone asks a direct question in a thread: will the migration finish this week, is the client likely to renew, did the bug cause the data loss. You have a view, but you are not certain. So you write “I think it should probably be fine, but it’s hard to say,” and send it.

That sentence protects you and helps nobody. The reader cannot tell whether you are 90 percent sure or 40 percent sure, so they either treat it as a yes, which turns your guess into a promise, or treat it as noise and ask someone else.

Most advice about hedging tells you to cut it. That is right when you actually know the answer, and sounding confident in work emails is mostly a matter of deleting “I think” and “hopefully” from sentences that do not need them. This post is about the other case: when the uncertainty is real and the reader needs to know its size.

Vague Hedges Hide The Number The Reader Needs

“Probably,” “should be,” “might,” and “it depends” all sound like information. They are not, because different readers decode them differently. One person reads “probably” as a near certainty; another reads it as a coin flip with a polite face.

Professions that live on uncertain judgments have noticed this. The US intelligence community’s analytic standards, ICD 203, require analysts to express likelihood using a fixed scale of terms, running from “remote” up to “almost certain,” instead of free-form phrasing. The same directive keeps two ideas apart: how likely something is, and how much confidence the analyst has in the sources behind that judgment.

You do not need a formal scale for a Slack reply, but both moves carry over: say how likely, and say what that rests on.

State How Sure You Are, Then Why

A useful uncertain answer has three parts. Your best answer, stated plainly. How sure you are, in words the reader cannot misread. And what the judgment is based on, so they can weigh it themselves.

Before:

I think the migration should probably be done by Friday, but there are a few unknowns so it’s hard to say for sure.

After:

Best guess: the migration finishes Thursday. I would put that at about 70 percent. The risk is the archive tables, which we have not tested at full size yet. If those run slow, it slips to Monday.

The second version is no more certain than the first. It is honest about the size of the uncertainty and names the one thing that would change the answer. A manager can plan around “70 percent, Monday as the fallback.” Nobody can plan around “hard to say.”

Numbers are not always appropriate. In a client email, “70 percent” can read as false precision or invite haggling. Plain words work if you pick them deliberately and use them consistently: “very likely,” “more likely than not,” “a real chance it slips,” “unlikely.” What matters is that you chose the word, rather than letting it fall out of nervousness.

Separate What You Know From What You Assume

Some facts in a message are confirmed, some inferred, and one or two are guesses. Mixing them makes the whole thing feel shaky.

Before:

It looks like the outage was probably caused by the config change, and it might have affected some customers, maybe around 200 or so.

After:

Confirmed: the outage started two minutes after the config change went out, and rolling it back fixed it. Likely but not yet verified: the change was the cause. Unknown: how many customers were affected. The logs give an upper bound of about 200, and we will have the exact figure by 3pm.

Labelling each claim lets the reader act on the confirmed parts immediately and wait on the rest. It also protects you later, and it fits naturally in a project blocker update or an incident note.

Always Say When You Will Know More

“I’m not sure” without a next step reads as a shrug. “I’m not sure yet, and I will know by Wednesday after the load test” reads as someone handling it.

The update point does two things. It tells the reader when to stop waiting, and it commits you only to reporting back, not to the answer itself. That is the same principle behind giving a time estimate without overpromising: name the checkpoint, not a guarantee.

When Not To Hedge At All

If you know the answer, state it. Habitual “I think” in front of checked facts trains readers to discount everything you write, including the times you really are unsure. If you are making a decision, own it, and put the risk in its own sentence rather than softening the decision.

If the honest answer is that you have no basis for a view, say that cleanly: “I do not know, and I am not the right person to guess. Priya owns the billing system.” That is more useful than a hedged opinion nobody should rely on.

Keep The Calibration Through A Rewrite

This is where AI rewriting tools need watching. Asked to make a message “more professional” or “more confident,” a model will often delete hedges wholesale, which turns “about 70 percent, Monday if the archive tables are slow” into “the migration will be complete Thursday.” The prose improves and the message becomes false. The opposite happens too: a model asked to soften may sprinkle “potentially” over confirmed facts.

A Wrivio Context for uncertain updates could say:

Rewrite this as a clear, professional work update. Keep every statement of likelihood exactly as strong or as weak as written: do not turn a guess into a commitment or a confirmed fact into a possibility. Keep labels such as confirmed, likely, and unknown. Remove filler hedges that do not carry meaning, such as “just” and “I feel like.” Keep every name, date, figure, and percentage exactly as written. Do not add new commitments or dates.

Press Ctrl+Shift+Space over Slack, Teams, or Outlook, paste your draft, and read the result against what you wrote before sending. The single thing to check is whether any sentence got more certain than you are. If the update involves an incident, customer counts, or anything you would not want leaving the building, switch to the Local engine so the text stays on your machine during the rewrite. For a fuller checklist, see how to review AI rewritten text.

Common Questions

Is it unprofessional to say “I don’t know” at work?

No. Saying you do not know, with a plan to find out and a time you will report back, reads as competent. What reads as unprepared is a vague hedge that hides how much you actually know.

Should I use percentages to express uncertainty in emails?

Use them with colleagues who plan around them, such as a manager or a project lead, because a number removes ambiguity. With clients or senior stakeholders, plain words like “very likely” or “a real chance it slips” usually land better and avoid false precision.

How do I hedge without sounding weak?

Hedge the specific claim that is uncertain, not the whole message. State your best answer plainly, give one sentence on how sure you are and why, and name when you will know more. Confidence comes from structure, not from removing every qualifier.

What is the difference between likelihood and confidence?

Likelihood is how probable you think an outcome is. Confidence is how solid the evidence behind that judgment is. You can think something is likely while having weak evidence, and saying both separately gives the reader a truer picture.

Download Wrivio for Windows to tidy uncertain updates without losing how sure you actually are.