A bad FAQ is a wall of corporate questions nobody asked. A good FAQ is a conversion tool: it pre-empts the five objections that almost stop the purchase. The prompt has to mine the questions buyers actually type, not invent flattering ones.
One ground rule first, because it changes the goal. Google fully retired FAQ rich results in 2026 (they stopped appearing on May 7, 2026, and Rich Results Test support was dropped that June), so a FAQ section is no longer a shortcut to expanded blue-link real estate. Where FAQs still pay off: on-page objection handling that lifts conversion, and answer-engine visibility. ChatGPT, Perplexity, and Google AI Overviews all extract question-answer pairs that lead with a direct answer, so a substantive FAQ remains one of the higher-leverage things you can add to a page. The prompts below are tuned for that reality.
TL;DR
- Mine real questions (support tickets, search queries, sales calls). Don’t invent flattering ones.
- Lead every answer with the answer; add one sentence of context after. Cap at 30-50 words.
- Order by buyer journey on landing pages; by frequency in help centers.
- FAQ schema no longer earns a Google rich result, but the content still wins AI-search citations and handles objections in-page.
- Paste a sanitized ticket export into a 1M-token model (Claude Opus 4.7, Gemini 3.1 Pro, or GPT-5.5) to cluster questions at scale.
Who this is for
SaaS founders writing landing pages, e-commerce sellers building PDPs, support leads turning tickets into a public FAQ, and content marketers chasing AI-search citations. If you’re writing a glossary or a full knowledge base, that’s a different format; these prompts are for the short, high-intent FAQ that sits below your pricing or product copy.
Which model to run these in
All 12 prompts work in any current chat model, but the ticket-mining ones (#1, #12) reward a big context window. As of June 2026, Claude Opus 4.7, Gemini 3.1 Pro, and GPT-5.5 all carry a 1M-token standard context, so you can paste a sanitized export of several hundred tickets in one shot instead of summarizing first. For the short answer-writing prompts (#2-#11), the workhorse tiers (Claude Sonnet 4.6, GPT-5.5 Instant) are faster and cheaper, and the quality gap is negligible. If you’re writing the answer copy by hand and only want the AI to polish it, see the companion FAQ writing prompts.
Prompt anatomy
Every FAQ-section prompt should carry six elements:
- Audience: one specific reader, not “anyone interested in X”.
- Goal: one outcome (read, click, sign up, agree, share).
- Voice: brand or author voice with 2-3 anchor adjectives.
- Constraints: word count, banned phrases, must-include facts.
- Format: paragraph, bulleted, headed sections, table.
- Examples: one or two samples of the tone you want; the strongest lever for matching voice.
Best for
- Landing-page FAQ that handles purchase objections
- PDP / pricing FAQ
- Onboarding FAQ for new users
- Turning support tickets into a public help article
- Answer-engine visibility (getting cited by ChatGPT, Perplexity, AI Overviews)
12 copy-ready prompt templates
1. Mine real questions from support tickets
I have these support tickets: {ticketLog}. Cluster into themes and identify the top 8 questions actually asked. For each: (a) the question rephrased to user voice, (b) clearest 2-sentence answer, (c) frequency rank. Skip questions only asked by one person.Variables to swap: ticketLog
2. Objection-handling FAQ
For my product `{productName}`, generate 6 FAQs that handle real purchase objections. For each: (a) the objection in user voice ("Will this work for a team of 2?"), (b) honest 2-3 sentence answer that addresses the objection, (c) optional small reassurance (free trial / refund). No "great question!" filler.Variables to swap: productName
3. Pricing-page FAQ
Generate 8 pricing-page FAQs covering: (1) billing cycle, (2) team seats, (3) annual discount, (4) refund / cancellation, (5) custom plans, (6) tax / invoicing, (7) usage limits, (8) plan downgrade. Each answer ≤ 35 words. No "contact sales" as the only answer.4. Ordering FAQs by user journey
Reorder these FAQs by where they arise in the buyer's journey: awareness → consideration → decision → after-purchase. Output as four grouped sections. If a question fits multiple stages, place it in the earliest one.5. Answer compression
Rewrite each of these FAQ answers in ≤ 30 words. Lead with the answer, then a sentence of context. Cut: "We understand…", "At {brand}, we…", and any apology that delays the answer.6. Answer-engine-aware FAQ
Write 6 FAQs that match real search queries for `{topic}`. Use the actual question form ("How do I…", "What is the difference between X and Y…"). Answers ≤ 50 words, factual, lead with the direct answer, no marketing language. Optimize for extraction by AI answer engines (ChatGPT, Perplexity, Google AI Overviews).Variables to swap: topic
FAQPage schema markup still helps machines parse these (Google parses it; Bing and RAG crawlers index it), but as of June 2026 it no longer earns a Google rich result. Add the schema for AI parsing, not for a SERP snippet.
7. Onboarding FAQ
New users of `{productName}` ask these 6 questions in week 1. Generate questions + answers. Answer style: walk them to the action ("To do X, go to Y → Z"). Include one link per answer.Variables to swap: productName
8. Trust-building FAQ
Write 5 FAQs that address trust concerns: data security, vendor lock-in, who built the company, support hours, change frequency. Answers should be specific (cite SOC2 status, list integrations to export) — vague answers reduce trust.9. Refund / cancellation FAQ
Write the 3 refund / cancellation FAQs we don't want to write: (1) Can I cancel anytime? (2) Will I get a prorated refund? (3) What happens to my data after cancellation? Answer plainly. Don't bury the answer in policy speak.10. “Compared to X” FAQ
Write 3 FAQs comparing `{product}` to known competitors `{compA}`, `{compB}`. Answer honestly — name the trade-off where the competitor is better, then the case for us. Reads as trustworthy, not as sales pitch.Variables to swap: product, compA, compB
11. FAQ rewrite for plain English
Rewrite these FAQs at a 7th-grade reading level. Replace jargon: "leverage" → "use", "robust" → "reliable", "solution" → drop or replace with the actual thing. No corporate hedges.12. FAQ audit
Audit this FAQ section: (1) Which questions answer themselves (drop), (2) Which contradict the product page, (3) Which buries an answer, (4) Which would lose trust if a skeptic read them. Output: keep / rewrite / drop per item.Common mistakes
- Inventing flattering questions (“How does your team innovate?”) nobody asked.
- Burying the answer in 3 paragraphs of context.
- Vague trust answers (“we take security seriously”).
- Pointing to “contact sales” as the only path.
- Not ordering by user journey.
- Writing in third-person corporate voice when users speak first-person.
- Skipping refund and cancellation because the topic is uncomfortable.
How to push results further
- Lead each answer with the answer; put context after.
- Cap answers at 30-50 words. Anything longer becomes its own help-center article.
- Use the user’s voice in the question, matching how they’d actually phrase it.
- Audit support tickets monthly to refresh FAQ topics.
- Add FAQPage JSON-LD for AI parsing, but don’t expect a Google rich result; that path closed in May 2026 (Google’s FAQPage docs now note the deprecation).
- Be honest about trade-offs. Readers reward it, and answer engines tend to cite the page that names the catch.
- Link to deeper help articles for the 10% who need more.
FAQ
- How many FAQs is too many?: 8-12 on a landing page. More belongs in a separate help center.
- Should I sort by importance or by journey?: By journey on landing pages; by frequency in help centers.
- Does FAQPage schema still help SEO?: Not for a Google rich result; that was fully retired in May 2026. The markup is still worth adding because Google parses it, and Bing plus AI-search crawlers use it to understand and cite your answers.
- Can AI mine my tickets?: Yes. Strip PII, then paste a sanitized export into a 1M-token model (Claude Opus 4.7, Gemini 3.1 Pro, GPT-5.5) and run prompt #1.
- Should I include negative reviews as FAQs?: Sometimes. Reframe the complaint as a real question and answer the trade-off honestly.
- When should I retire an FAQ?: When ticket volume on that question drops below a threshold (for example, under 5 a month).
Related
- FAQ writing prompts
- Landing page copy prompts
- Sales copy prompts
- CTA prompts
- Writing & Copywriting Prompts hub
Tags: #Prompt #Writing #FAQ #Landing page