How can I practice making a recommendation from incomplete information in an interview?

Treat the missing information as part of the task, not as a puzzle with one hidden answer. Use a short scenario and organize your response into four parts: known (what the prompt says), unknown (what could affect the choice), assumption (what you will provisionally take to be true), and recommendation (what you would do now, why, and what could change your mind).

For example, suppose a team must choose between releasing a feature on Friday or delaying it for another review. The prompt says it is nearly ready but gives no details about the remaining issue, the cost of delay, or who is affected. Do not invent those details. You might say: “I don’t know the impact of the remaining issue. For now, I’ll assume it is minor and that delaying has a meaningful cost. I’d recommend a limited release only if the team can contain the impact; otherwise I’d delay and clarify the risk first.”

That makes your assumption visible and keeps the recommendation conditional instead of presenting a guess as fact.

Ask a few focused questions if the format allows it. Prioritize questions whose answers could change your choice; you do not need every detail. For the example, you could ask, “What is the remaining issue, and who could it affect?” or “What happens if we delay?”

Sort possible questions into three groups: decision-changing, helpful but not essential, and unlikely to affect the choice. Ask the first kind first. If answers are unavailable—or questions are not invited—briefly name the important unknowns, state your assumptions, and proceed. This keeps the conversation moving without pretending the gaps do not exist.

A clear conditional answer is still a decision. Try this structure: “Based on [known facts], I recommend [action]. I’m assuming [important assumption]. This is the best next step because [main trade-off]. I would change my recommendation if [specific new information]. Before acting, I would confirm [one key point].”

For the release example: “The feature is nearly ready, but one issue remains. I recommend delaying the full release until the team understands that issue. I’m assuming its impact is not yet known; without that information, a full release would go beyond what the facts support. A short review is a proportionate next step because it could clarify the concern while keeping a later release possible. I’d recommend proceeding sooner if the issue is confirmed to be minor and contained. First, I’d confirm who could be affected and what the delay would cost.”

You have chosen a next step without claiming certainty about the final outcome.

After each attempt, write down your answer and review its reasoning—not whether you guessed an imagined “correct” answer. Check whether you separated facts from assumptions, chose useful questions, made a recommendation, and named a specific condition that could change it.

Then change one feature of the scenario, such as making delay more costly, and try again. Your recommendation should respond to that changed fact; if it does not, revisit your reasoning. Keep prompts with different gaps, such as unclear priorities, missing timing, or uncertain consequences, so you are not just rehearsing one script.

You cannot know exactly what an interviewer expects from an unspecified scenario. Focus on making your reasoning understandable and open to revision.