Choose the public storefront

Use the canonical website customers visit. Check language, country, and the difference between the main domain and regional storefronts. Only submit websites you are authorised to connect. A password-protected administration page is not a suitable knowledge source.

Check representative pages

After the crawl, inspect several product pages and at least one delivery, returns, and contact page. A successful crawl count does not establish that every important fact was captured. Some pages may contain little readable content or require a browser interaction.

Connect sources to outcomes

Ask questions that require the pages you inspected. If the chatbot cannot answer, check both the source content and indexing status. Avoid immediately starting another full crawl: the initial indexing may still be running.

Check coverage beyond the main domain

Product facts, delivery dates, and support policies sometimes live on separate help sites or subdomains. Verify that the exact page containing the information is connected and readable. A link to another website does not guarantee its contents were imported.

Connect policy pages, help sites, and subdomainsMake product specifications useful to the AIUnderstand website and catalogue import progress

Technical details and methodology

For the collection steps, eligibility rules, calculation units, and limits behind this workflow, use the implementation-reviewed technical guides below.

Website discovery, crawling, and knowledge indexing