How to explain uncertainty clearly at work

The listener needs to know which part they can use now and which part would be a guess. You can be direct about a completed task while keeping its untested result open.

Completion and performance are different claims

Consider this fictional situation:

• A migration is complete.
• The overnight job must finish within an agreed target window.
• No comparable run on the new setup has finished yet.
• The team will review the first run tomorrow morning and give an update at nine.
• For this example, the agreed operating plan stays in place until that review.

These are authored facts for rehearsal, not results from a Ptichi user or workplace study.

A vague update might sound like this:

The migration looks okay. We still need to check a few things, so we're not completely sure yet. Hopefully we'll know more tomorrow.

“Not completely sure” spreads doubt across the whole message. It hides both a confirmed completion and the exact result that nobody has observed yet.

Keep those claims separate:

The migration is complete. We do not yet know whether the overnight job will finish within the target window. Tomorrow morning's run will give us the first comparable result. I'll update you at nine, even if it is still unresolved. Until that review, we're keeping the agreed operating plan.

The listener can now recover the completed work, the open outcome, the next evidence and the current decision. Keeping the plan is part of this fictional scenario; uncertainty alone does not tell every team to wait.

Use four recoverable parts

Before speaking, write a short factual note:

1. Known: what is already established?
2. Uncertain: what exactly remains open?
3. What could change it: what observation, check or decision could change the answer?
4. Next update: when should the listener expect another update or decision?

Add the practical consequence when it matters: what can proceed now, what needs a condition, and who owns that decision? You may be giving information to the decision-maker rather than choosing for them.

For a confirmed delay, the changed commitment needs to be explicit. A risk update also needs a possible adverse consequence and an action trigger. This guide helps when the unresolved item itself needs explaining, including situations where neither a delay nor a risk has been established.

Name the uncertainty, not just the mood

Different open questions require different information. “We're not sure” does not tell the listener what is actually unresolved.

What is uncertain?	What to make explicit
Outcome	What result is still open, and what evidence will resolve it?
Date	What condition controls the date, and when will that condition be known?
Cause	What has been ruled out, what is still being tested, and what would count as evidence?
Estimate	What assumptions or range matter, and what new information could move the estimate?
Decision	What is known well enough to decide now, and what would justify waiting?

Do not invent numerical confidence merely to make the answer look precise. If you have an actual measured range or probability, preserve its source and assumptions. If you do not, describe the condition that keeps the answer open.

A checkpoint can arrive before the answer

“I'll update you at nine” is a commitment to communicate. It does not mean “the uncertainty will be resolved by nine.”

If the run is incomplete, the update can say that, identify the evidence still missing and name the next useful check. Do not promise a finished answer simply because a meeting is on the calendar.

Sometimes there is no useful next check within your control. Say what dependency you are waiting for and what you can do meanwhile. A guessed checkpoint would add another uncertainty.

Rehearse the boundary between fact and outcome

Use one real, non-confidential open question or the fictional migration facts above. Keep your factual note short. Optional playback can use a recorder you already have; the Ptichi website does not record audio.

Take A

Answer naturally out loud:

“The migration is done. Can we rely on the overnight job now?”

Keep the scenario's facts. Speak before polishing a paragraph.

Listen

On one playback, notice:

• Can you distinguish the completed migration from the unobserved overnight result?
• Did “probably,” “hopefully” or “not sure” replace the actual open question?
• Can you hear which check could change the answer?
• Did you promise an update, or accidentally promise that the result would be known by then?

Leave filler words, pitch and general confidence aside for this pass. You already know the facts, so self-listening reviews the wording; it does not establish what an unfamiliar listener understood.

Change one thing

For Take B, make one change:

State the confirmed fact and the unresolved result as separate claims.

For example: “The migration is complete. We haven't observed its overnight result yet.” Preserve the facts and any evidence, plan or update point you included in Take A. Note other omissions for a separate pass. Do not also invent a probability, revise the plan or optimize every sentence.

Take B

Answer the same question again, using your own wording. The uncertainty should belong to the overnight result rather than to every part of the update.

Compare

Which take makes the known fact and open outcome easier to distinguish? Did Take B preserve the evidence and update point, or sound cleaner by deleting them?

If a willing colleague listens, ask what is confirmed, what remains open and when another update is due before showing the factual note. Keep their original response, including uncertainty or missing details. Understanding your account and agreeing with the operating decision are separate checks.

A colleague who knew the situation or heard Take A brings that knowledge to Take B. New wording does not erase it, and a later correct answer does not isolate the effect of your one change.

Transfer

Try this different fictional situation: two offices have sent feedback on a new workflow. Other offices have not replied. You can describe the responses received, but you do not yet have the other offices' views. Collection closes Friday; the wider summary is due Monday. The immediate decision is whether to fix a problem reported by one office before that wider summary.

Explain what the feedback establishes and what it leaves open. Use new wording, then say what evidence would change the wider answer and what the decision-maker can consider now.

This is another practice attempt. It does not prove lasting learning or success in a later conversation.

Useful caution earns its place

Make caution specific. “Treat this estimate cautiously” leaves the listener to find the reason. “This estimate assumes the supplier sends the file Friday” names a condition they can check.

Preserve measured uncertainty. If an estimate comes with a range or probability, keep its source, assumptions and relevant limits. Do not replace it with vague reassurance. If no defensible number exists, say what is open and why.

Keep decisions separate from predictions. A team may choose a reversible next step while an outcome is unknown. That choice does not establish the outcome, and this speaking guide does not decide whether the step is appropriate.

Evidence and our interpretation

The UK Government Analysis Function's guidance on communicating uncertainty asks statistics producers to explain important limitations and what users can conclude. Published in 2018, it is currently marked under review. It is statistical guidance, not a test of this speaking exercise.

Our read: for an ordinary work update, make the boundary visible where the listener might otherwise mistake a completed action for a confirmed result. The factual note and one-change rehearsal are Ptichi's editorial adaptation. We do not know whether this exercise improves a later workplace conversation.

When a short update is insufficient

An uncertain cause may remain uncertain even after a good explanation. Rehearsal lets you inspect the wording; it cannot produce missing evidence.

Formal statistical reporting, engineering acceptance or regulated decisions may require documented methods, quantitative uncertainty and review. Keep those processes and the evidence they need. A spoken outline cannot replace them.

If a decision cannot wait, explain what is known, what could go wrong and what options remain. The right action depends on the stakes and authority, rather than on how neat the update sounds.

Continue with Ptichi

Bring one open question to the free pre-meeting rehearsal, which works without an account. Keep the confirmed fact, unresolved result, next evidence and next update in your notes. Practise separating the first two, then try another situation with different wording.

The web exercise and guide are available now. Optional playback uses your own recorder; Ptichi's desktop application remains in development.