Build a Knowledge Base for an Online Store Selling in Europe

To build a knowledge base for an online store selling in Europe, start with questions customers actually ask. Publish answers that apply to many shoppers, route individual order and account questions to private support, and test whether the published answers are findable and still accurate.

Start with support questions, not imagined FAQs

Review inbox messages, chats and searches on your existing FAQ page. Record the customer’s wording, the answer given, the source used to verify it, whether the answer can be public, and who owns the content. Keep pre-purchase questions about products distinct from post-purchase questions about a specific order.

Customer question	Reusable public answer	Private support path
“Do you deliver to my country?”	Current destinations and where to check shipping options.	Questions about a specific address or shipment.
“How do I return this item?”	Verified return steps, conditions and where to read the store’s policy.	Decisions about a particular purchase or return request.
“Where is my order?”	How to find tracking information and what to do if it is missing.	Checking the customer’s actual order or delivery problem.
“Can you change my account email?”	Verified self-service instructions, if your store offers them.	Account changes or access problems requiring verification.

Prioritise questions with stable answers you can verify. A repeated question is not necessarily a good public article if the useful answer depends on private order details.

Decide what can safely become a public article

Use this decision rule: publish the general process, not the customer’s case. If a resolved chat suggests an article, rewrite the general question and check the answer against your current policy rather than copying the exchange.

This is an editorial check, not legal advice about returns or GDPR. If you use conversations to identify article ideas, treat reviewing chat transcript retention for your store as a separate privacy task.

Organise articles around shopper intent

Group the questions you collected by the task the shopper wants to complete. Combine topics when the same answer serves them; separate topics when the next action changes. For example, finding a tracking link and reporting a delayed parcel need different instructions.

1. Use the customer’s words. Title an article “How do I track my order?” rather than “Fulfilment status”.
2. Separate journey stages. “Do you ship this product to my country?” is a pre-purchase question; “Where is my parcel?” concerns an existing order.
3. Check placement. Put sizing guidance near the relevant product and make return instructions findable from your help area and relevant post-purchase messages.
4. Make the limit clear. If an article cannot explain what happened to an individual order, show a private route to your team.

Before adding a new page, check whether improving an existing answer would make it easier to find. That is a useful first step when reducing support tickets with self-service.

Write an article that tells the customer what to do next

Answer one question per article. Start with the direct answer, give the actions in order, and say when the customer needs private help. Use this return-article template as a draft, not as a ready-made policy; replace every bracketed field with wording verified for your store and its destinations.

Before publishing, follow the instructions as if you were a customer. If “request a return” leads to a page that does not explain how to make the request, the article is unfinished.

Test search and the handoff before launch

A shopper may search for “Where is my order?” while your team calls the topic “fulfilment status”. Test the shopper’s phrase, “delivery update” and wording taken from real enquiries. If a search produces no useful answer, check the existing article’s title and wording before creating a duplicate.

askVela’s public help centre has search, and its knowledge base can also provide the source for AI answers. Test what the customer sees rather than assuming that publishing an article completes the job.

Keep answers current and identify gaps

Assign an owner to each policy-sensitive article and review it when the underlying process changes:

Compare recurring support questions with searches that produced no useful answer; article views alone cannot tell you whether a problem was resolved. In askVela, unanswered-question collection surfaces questions its bot could not answer, while resolved chats can become draft help articles. Review every proposed article before publication: verify the policy, remove customer-specific details, test the search phrase and approve the final wording.