Find questions worth answering.

Read the conversation for the task someone is trying to complete. Look for where they got stuck and what an answer would need to explain.

• Look for repeated questions. One comment can suggest a topic. The same question in separate conversations gives you more reason to investigate it.
• Keep the source context. Record who asked and what they had tried. Include the decision they were facing.
• Group questions by need. People can describe the same problem in different words. Check what they need answered before grouping their comments.

Check how readers would find a proposed page and whether your site already answers the question. Make sure your team knows enough about the topic to give a useful answer.

• Check search demand. Read the queries to understand what people expect to find. Search volume alone cannot tell you whether the answer helps.
• Review existing pages. Update a page if it already covers the question.
• Choose a format. The question might need a full guide or just a section added to an existing page.

Use our content gap analysis guide to compare a question with your existing pages and decide what needs to change.

Our guide to turning customer conversations into content shows this decision through an example.

To collect questions from public threads first, read our guide to Reddit keyword research.

After publication, search queries can show which questions bring people to the page. Use them to spot answers that need work. Recorded AI appearances can help too, though they cannot account for every response an AI system gives.

• Read the queries. Check whether visitors are looking for the answer your page gives.
• Review the page. If readers expect a different answer, revise the content or how you describe it.
• Treat visibility as an observation. Google and AI systems choose their sources. A tool cannot guarantee those choices.