FAQ from customer questions: collect, sort, answer

Why your inbox is the best source

Every week customers ask the same things: where is my parcel, can I return this, do you deliver abroad. Each answer costs time, and every customer who has to ask was missing something on your website. The messages show you what that is, in your customers' own words. An FAQ built from them answers real questions, not the ones you guess people might have.

From the inbox to the FAQ

Step 1: collect

• Choose a fixed period, for example the last one to three months, so that one busy week does not decide everything.
• Use all channels: e-mail, chat, contact form, notes from phone calls, comments and reviews.
• Export the messages as text or CSV. You need the customers' questions, not your team's replies.

Step 2: remove personal data first

Customer messages are full of personal data. The GDPR defines it as any information relating to an identified or identifiable person, for example a name, an identification number or an online identifier (Art. 4(1)). For an FAQ you need the question, not the person. This fits the principle of data minimisation: personal data must be limited to what is necessary for the purpose (Art. 5(1)(c)).

• Remove names, e-mail addresses, phone numbers, postal addresses and order or customer numbers before you analyse or share the messages.
• Delete greetings and signatures. They often contain the name.
• Leave out messages with health data or other special categories: racial or ethnic origin, political opinions, religious or philosophical beliefs, trade union membership, genetic and biometric data, sex life or sexual orientation. The GDPR prohibits processing these data unless one of the exceptions in Art. 9(2) applies.
• Never copy a real message into the FAQ. Write the question in neutral words.

Step 3: group and count

Give each message a topic, such as delivery status, returns, payment or sizes. Merge topics that mean the same thing. Then count the messages per topic. The count shows you where to start. A rare question can still matter, for example one about safety, but most of your effort belongs at the top of the list.

Step 4: write the answers

• Use your customers' words in the question. They ask 'Where is my parcel?', not 'Shipment tracking information'.
• Give the answer in the first sentence, then the details.
• Write only facts you know and can keep: delivery times, prices, deadlines. If a fact is missing, find it out before you publish.
• Keep it clear and short. The W3C recommends simple language, short sentences and lists.
• Link to the place where the reader can act, for example the tracking page or the returns form.

Google asks site owners whether a reader will leave a page feeling they have learned enough to reach their goal. That is a good test for every FAQ answer too. Write the FAQ for your customers, not to attract visits from search engines.

Step 5: find the gaps

Compare your list of topics with the FAQ you already have. Three kinds of gaps are common:

• Frequent questions that your FAQ does not answer at all.
• Questions it answers, but in words your customers do not use, so they do not find the answer.
• Answers that are outdated or too vague, so customers ask anyway.

FAQ page or better product page?

Not every frequent question belongs in an FAQ. If many people ask about one product or one step, the answer is often missing exactly there. A size chart on the product page helps more than a size question on an FAQ page that few people open. Use the FAQ for questions about your shop as a whole, and improve the page itself where the question arises.

Step 6: keep it current

• Repeat the count every few months and compare: does a topic get fewer messages after you answered it?
• Change the FAQ when prices, delivery times or rules change. An outdated answer causes new questions.
• Give each answer an owner who knows when the facts change.
• If you add FAQ markup to the page, it must say the same as the visible text. Our guide on FAQ and structured data explains how.

Work with Do4U

The tool FAQ from support messages reads many customer messages at once, pasted as text or as a CSV export. Before the AI reads anything, the code removes e-mail addresses, phone and order numbers, addresses and names in greetings and signatures. Do4U groups the questions into topics, and the code counts the messages behind each topic. For each topic you get the question in your customers' words and a draft answer that only uses facts from the messages or from your FAQ page. Where a fact is missing, the draft shows a gap for you to fill in. If you give the address of your FAQ page, Do4U also shows which questions it already answers, with a quoted sentence as proof, and which frequent questions are missing. You can pass the drafts on to the FAQ and schema markup tool.

Do4U builds the draft, you check and decide. The code cannot find every personal detail, so do not paste health data or other special categories. Fill in every gap before you publish an answer.