Use when you want to see which questions in a topic your site answers and which it does not.
You are mapping question coverage for a topic so we can see where assistants have nothing of ours to cite.
Topic: {{TOPIC}}
Our audience: {{AUDIENCE}}
Pages we already have, with URLs and one-line summaries: {{EXISTING_PAGES}}
Steps:
1. Generate the question set a buyer works through on this topic, from first awareness to post-purchase. Aim for 35 to 50 questions, phrased as a person would type them.
2. Map each question to the existing page that answers it, if any.
3. Score the answer quality 0 to 3 where 0 is not covered, 1 is mentioned in passing, 2 is answered but buried, 3 is answered directly and self-containedly.
Output:
| Stage | Question | Best matching page | Score | What is missing |
Then give:
- The ten highest value gaps, ranked by how directly they precede a buying decision.
- Five questions currently answered across multiple pages that should be consolidated into one.
Constraints:
- Judge coverage only from {{EXISTING_PAGES}}. If a summary is too thin to judge, score it "unknown" rather than guessing.
- Do not invent questions that only make sense if our product exists.
- Do not recommend a new page where an existing page could answer the question with an added section. Say which.
Replace each placeholder with your own detail. The more specific you are, the less the model invents.