Evaluate clarification when a reader’s question is ambiguous

Find the detail that changes the answer

A question such as how do I export may omit the object, format or product version. List which missing details materially change the documented steps. A good response can ask for that detail or state a clearly labelled assumption. It should not guess silently when the choice could produce the wrong instruction. Keep the expected clarification narrow enough that a reader can answer it easily.

Avoid an interrogation script

Not every brief question needs a follow-up. If the public documentation has one supported export workflow and the question is otherwise clear, answer directly with its limits. Excessive clarification adds friction and can make a simple task slower than ordinary search. Evaluate whether each requested detail is necessary for correctness rather than rewarding a chatbot simply for asking more questions.

Test the conversation, not only the first turn

Use a fictional exchange in which the reader supplies the missing version after a clarification. Check that the final answer uses that version rather than reverting to the default. Keep the conversation and its expected source together. Do not count a sensible first question as a completed success when the subsequent answer ignores the reply or blends instructions from different releases.

Record assumptions explicitly

When a response proceeds under an assumption, ask whether the assumption is visible and easy to correct. A reader should not need to infer which plan or version the assistant chose. Use these findings to improve source labels and examples as well as chatbot instructions. Ambiguity is sometimes a documentation design problem, and a clearer article may help both chat and non-chat readers.

• Record the missing detail, expected clarification and final answer after clarification; the first response alone cannot establish that the task succeeded.