Finding the questions you have not answered yet
Answer engines reward pages that answer specific questions. The bottleneck is almost never the writing — it is knowing which questions to answer. The best sources are not keyword tools; they are the places where people have already asked, in their own words, and someone on your side had to reply.
Keyword research returns phrases. Question research returns problems. The difference matters because an answer engine is matching an intent expressed as a sentence, and a two-word phrase with a volume estimate next to it does not tell you what the person actually wanted to know.
Start with the questions you are already being asked
Every organisation sits on a large, well-qualified, entirely unused corpus of real questions. Nobody mines it because it does not arrive labelled as research.
The practical method is unglamorous: take a quarter of support tickets and sales notes, group them by the underlying question rather than by category, and count. The top ten groups are your content plan, already validated, already phrased the way your audience phrases things.
People Also Ask, read as a tree rather than a list
The related-questions block in search results is the most direct public signal available about which questions a search engine associates with a topic. Its useful property is that it expands: opening one question adds more, so the block behaves as a tree of related intent.
- Search your head term and record the initial related questions.
- Expand each one and record the questions that appear beneath it, which are conditioned on the first.
- Go two levels deep, no further — beyond that the questions drift away from the original topic.
- Group the results by intent rather than by wording, since several phrasings usually share one answer.
- Note which of your existing pages could answer each group, because most of the work is often adding a section rather than writing a page.
The other sources worth an hour
- Autocomplete — type your head term followed by each letter of the alphabet, and by why, how, can, should, is, does. The suggestions are real query prefixes
- Search Console query report — filter to queries containing question words, then look for ones with impressions and almost no clicks. Those are questions you appear for and do not answer
- Community threads — forum and discussion sites where practitioners ask each other. Long answers with heavy agreement mark questions that are genuinely hard
- Review text for competing products — complaints describe the problems buyers have; each one is a question someone asked before purchasing
- Your own analytics on exits — a page where people leave without continuing usually answered less than it promised
The Search Console technique deserves emphasis because it is the only one that uses your own measured position. A query with a thousand impressions and eleven clicks is one where you are visible and unconvincing — which is a much cheaper problem to fix than one where you are invisible.
Which questions are worth answering
Not all of them. A question list is only useful once it is ordered, and volume is the least useful ordering criterion available.
The third row is the constraint people ignore. A question you can only answer generically produces a page indistinguishable from the twenty that already exist, and there is no mechanism by which it becomes the cited one. Answering fewer questions from actual experience beats answering more from research.
Telling a question from a keyword
Some strings that look like questions are keywords with question words attached, produced by tools rather than people. They are worth recognising, because writing for them produces content nobody asked for.
The right-hand column is longer, less frequent, and far more answerable. It is also closer to how someone types into an assistant, which is the shift that makes this style of research more valuable than it used to be.
Run it as a standing habit, not a project
- Monthly — scan support and sales for questions that appeared for the first time
- Quarterly — re-expand the related-questions tree for your main topics; it changes as the topic does
- Quarterly — re-run the impressions-without-clicks report and fix the top few
- On every enquiry — ask whether the site should have answered it, and file it if the answer is no
The last habit is the one that compounds. A team that treats each unanswered enquiry as a content defect converges on a site that answers the real questions, without ever needing a research project to find them.
Questions
Where do the best content questions come from?
Your own support queue, pre-sales calls, site search logs and email enquiries — places where a real person asked in their own words and someone had to reply. They are pre-validated, phrased better than anything you would invent, and unavailable to competitors. Keyword tools return phrases; these return problems.
Should I create a separate page for every question?
No. Near-identical thin pages compete with each other, dilute the topic and read as automated output. Group questions that share an answer and write one substantial page per group, with each question as a section heading. A page answering eight related questions well outperforms eight pages answering one each badly.
How do I find questions I am already visible for but not answering?
Filter the Search Console query report to queries containing question words and sort for high impressions with very few clicks. Those are questions where you already appear and fail to convince — a far cheaper problem to fix than one where you are absent, because the visibility already exists.
Is search volume the right way to prioritise questions?
It is the weakest signal of the useful ones. Rank higher the questions asked by people who went on to buy, the ones your support team answers repeatedly, and the ones you can answer from genuine experience. A high-volume question you can only answer generically produces a page indistinguishable from the twenty that already exist.