CONTENT WRITING

FAQ section from real questions

Use when you have actual user or sales questions and want an FAQ that answers them rather than a keyword list.

faq-from-real-questions.md

Download .md

You are building an FAQ section from real questions, not from keyword tools.

Page topic: {{PAGE_TOPIC}}
Real questions collected from support, sales or search console:
{{RAW_QUESTIONS}}
What the page already covers in the main body:
{{BODY_COVERAGE}}
Facts I can confirm:
{{KNOWN_FACTS}}

Steps:
1. Deduplicate and cluster the questions. Show the cluster, the phrasings collected, and how many times each cluster appears.
2. Drop any cluster already answered well in the main body, and say which section answers it.
3. For the remaining clusters, write the FAQ. Each entry: the question in the wording real people used, then an answer of 40 to 70 words that starts with the direct answer in the first sentence.
4. Order the entries by frequency, then by how likely the answer is to change a buying decision.

Rules:
- Do not invent questions to fill the section. If only four survive, publish four.
- Where an answer needs a fact I have not confirmed, write [NEEDS CONFIRMATION: what] instead of guessing.
- Do not begin answers with restatements such as Great question or That depends.
- No em dashes. British spelling.

Fill in before running

Replace each placeholder with your own detail. The more specific you are, the less the model invents.

{{PAGE_TOPIC}}
{{RAW_QUESTIONS}}
{{BODY_COVERAGE}}
{{KNOWN_FACTS}}

Getting a better result

Search Console queries with impressions and no clicks are the best raw input here.
Keeping the original phrasing matters more than tidying it into grammatical questions.
Anything dropped in step 2 is a sign the body section needs a better heading.

Questions about this prompt

When should I build an FAQ this way rather than from keyword research?

When you have questions people actually asked, from support, sales or Search Console. The prompt deduplicates collected phrasings and counts frequency, so a keyword tool export gives it nothing to count. Queries with impressions and no clicks are the strongest raw input you can hand it.

What counts as usable raw questions?

The page topic, the raw questions in the wording people used, what the main body already covers, and the facts you can confirm. Keep the original phrasing, ungrammatical or not. Tidying questions into neat search phrases destroys the signal the clustering step is there to count.

What does it produce beyond the FAQ entries?

Clusters with their phrasings and counts, the clusters already answered in the body, then the surviving entries with forty to seventy word answers that lead with the answer, ordered by frequency and buying impact. What gets dropped in step 2 is worth reading. It usually means a body heading is unfindable.

What ruins an FAQ built this way?

Topping the section back up. It is told to publish four entries if only four survive, and adding invented questions to reach ten is how an FAQ turns back into a keyword list nobody reads. Every NEEDS CONFIRMATION marker also has to be answered by a person before the answer goes live.