TL;DR: Raise material bad news early enough for the decision owner to act, clearly separating facts from forecasts. Recommend a feasible next step and state what decision or support you need, even when the full diagnosis is still developing.
How to approach it
Use a real case where the timing of disclosure affected the available options. Explain what you knew when you reported it and what remained uncertain. Include your contribution to the problem if there was one, without making the whole account about your reaction.
A strong answer
Lead with the consequence: a cutover date is at risk, recovery may exceed its objective, or an acceptance condition has failed. Then state the evidence. “The rehearsal failed” is a fact; “we need another week” is a forecast that needs assumptions. Keeping those separate lets leadership act without mistaking uncertainty for concealment or a preliminary estimate for a promise.
Consider a fictional migration exercise. A cutover is scheduled in ten working days. The rehearsal reveals a data mismatch that has not been explained. The team estimates two to four working days to investigate and repair it, followed by three days for a complete rehearsal and review. That creates a five-to-seven-day technical path if the fix works on its first attempt, leaving limited time for another failure or a required approval.
Report the threat now. Waiting for a complete root cause uses the very time needed to choose a different cutover slot. Explain why the estimate may change: the mismatch could come from a mapping defect, missing source data or a write occurring during transfer. Each has a different repair and validation cost.
A concise fictional update could be: “The cutover is at risk because the rehearsal did not preserve the agreed data invariant. We have five to seven working days of estimated repair and retest under the current hypothesis. I recommend retaining the existing service and deciding on a new slot if tomorrow's investigation does not establish a bounded fix. I need the delivery owner to confirm how long the current service can remain supported.”
That update gives leadership an actionable dependency. It does not ask them to decide which SQL statement to run. If you can recommend a smaller scope, verify that it is operationally separable before offering it. Leaving half the data on each system can create synchronization work that is larger than the original migration.
Own the estimate error if you omitted concurrent writes or underfunded the rehearsal. State the corrective change, such as testing the transfer with representative writes before committing the next date. Avoid claiming that disclosure guarantees a favorable reaction; it enables an informed decision even when the conversation is difficult.
For client-facing delivery, identify who owns the external update and supply a shared factual record. Routine coordination should not delay required incident notification or an urgent reporting path. If the normal contact is unavailable, use the documented escalation route and make the uncertainty explicit.
Close the decision loop with the chosen plan, owner and next checkpoint. If the investigation disproves the initial hypothesis, update the forecast and explain what changed. Retain earlier estimates as historical context rather than silently rewriting a status report to look prescient.
What interviewers probe next
“Why report before you were certain?” Explain the consequence of waiting and the option that would disappear. You can report a credible risk without claiming that the worst outcome is certain.
“Leadership asks for a single date.” Give the best supported commitment or forecast with its conditions. If the evidence does not support a firm date, identify the next check that will narrow the range rather than inventing precision.
“The problem was your mistake.” Use the same evidence and decision structure, naming your action and the mitigation underway. Do not wait to repair it alone before informing people who can contain its effects.
Common mistakes
Burying the failed acceptance condition under a mostly green status report can leave readers with the wrong conclusion. Put the affected commitment in the headline.
Offering extra contractors without explaining which critical task they can shorten is an unsupported recovery plan. Show the resource dependency and onboarding cost.
Practice delivering the example in under a minute, then answer which assumption would most change the forecast.