FAQ Build & Review Kit

When this helps

You want to build an FAQ, but counting common questions and confirming each answer is accurate takes real effort

What it does

Builds a FAQ from inquiry records, emails, chat exports and internal manuals. It extracts questions, counts the most common ones, and masks names, emails, phone and order numbers. Each answer is tagged with its source (which manual or price list) and who confirmed it, then checked for internal jargon, staff names, mismatched numbers, and conflicting answers to similar questions. Only answers that pass become a ready-to-publish web page, plus an internal FAQ and Excel ledger. It also watches review deadlines, changes to source documents, and unanswered questions. Legally sensitive wording, like mail-order return notices, is flagged to check too — the lawfulness call is yours to make.

The fixes baked into this kit

1. People assume adding FAQ structured data will get it listed in searchGoogle's official search update log confirms that from May 7, 2026, search results no longer show the FAQ rich result. This kit skips structured data entirely and builds pages that are readable when opened and searchable by keyword instead.
2. Writing "we cannot accept returns" on a mail-order siteFour of eleven public FAQ pages reviewed had this wording. Under Japan's mail-order rules, returns must be accepted within 8 days of delivery unless a return policy is stated in the ad and on the final confirmation screen. For mail-order businesses this is flagged to check, and the kit asks whether that policy is posted on the site.
3. Treating "shipping fee" and "COD fee" as the same questionPhrasing words like "how much" or "how long" are stripped before comparing questions, so look-alike but distinct questions — shipping vs. COD fees, postpay vs. convenience-store-payment fees — are not flagged as conflicting answers.
4. Mistaking "free consultation available" for a price statement"Free" is excluded from price checks when it refers to a consultation, trial or estimate. "Same day" and "deadline" are only treated as firm promises when they appear in set phrasings.
5. Manual's "within 2 business days" turns into "within 2 days" in the answerNumbers in answers are compared against the exact wording in the source document's heading. Mixing up days vs. business days, or tax-included vs. tax-excluded, is flagged to check; a number with no source at all is flagged to fix.
6. Name, email and phone columns in inquiry spreadsheets leaking into the AIOnly the question and date columns are read from inquiry tables, and names, numbers and contact details left inside the text are masked. Quoted replies and email signatures are never read either.
7. Changing the shipping fee makes its answer vanish from the web page, hiding version differencesSwapping in an updated source document can drop an answer whose numbers no longer match from the web page. The kit flags it when a new version has fewer questions than the last one, and keeps the previous version's page for comparison.
8. Customer notices like "no need to confirm before your visit" mistaken for instructions to the AICustomer-facing notices that start with the polite prefixes "go-" or "o-" before phrases like "confirmation is not needed" are not treated as instruction-like text aimed at the AI.

What's inside

• A tool that pulls questions out of inquiries, emails and chats and counts them (never reads name, phone or email columns)
• A tool that imports manuals and price lists as sources, keeping every past version
• A tool that builds customer-facing and internal FAQs, each tagged with its source, who confirmed it, and a review deadline
• A check that flags internal jargon, staff names, mismatched numbers and legally sensitive wording
• A tool that builds a ready-to-publish web page plus a snippet to paste into your existing site
• A tool that builds an internal FAQ and an FAQ ledger (Excel)
• A tool that reviews an existing FAQ page you already have (read-only)
• Monitoring for review deadlines, changed source documents, and unanswered questions

The first three decisions

1. Your company name
2. A contact customers may be given (phone, email, inquiry page)
3. Whether you sell by mail order (online, phone or catalogue)

What you need

• No extra costClaude Code or CodexBoth CLAUDE.md and AGENTS.md are included
• No extra costNode.js 18 or newerOnly used to run the kit's tools. Nothing else to install
• No extra costExcel (or compatible software) and a browserUsed to view the ledger, web pages and results page

How to use

Just paste this into your AI.

I want to use the nice to have kit "FAQ Build & Review Kit". Follow https://nicetohave.app/en/g/NSTBU4G4-•••••••••• and set it up.

Works the same in Claude Code or Codex.
Read what this tells your AI to do, before you paste it

FAQ

Can it produce a foreign-language FAQ or translate it?

No. This kit only builds a Japanese-language FAQ. Foreign-language FAQs, translation and chatbot integration are out of scope.

Can it read image-only FAQs or text inside a PDF?

No. It can't read text inside images or PDF bodies, so copy the text into a .txt file before handing it over.

Can it draft reply emails to inquiries?

No, not this kit. Drafting replies is handled by a separate "Inquiry Reply Kit", which can use the FAQ this kit builds as its source.

Does it decide whether an answer is legally compliant?

No. It only flags things to fix or check, and what to verify. Whether something is actually unlawful is for you or a professional to decide.

How are review deadlines set?

Answers involving pricing, deadlines or legal wording default to a 3-month review cycle; everything else defaults to 6 months. These are deliberately cautious defaults your company can change.

Supporters

Once support opens, the people who backed this kit will be listed here.