Run a practical pre-launch check

Test a real customer journey#

Use the published storefront in a private browser window, rather than relying only on dashboard previews. Test the same language, market, and device your customers use. Keep a small list of expected answers and product results so changes can be checked consistently.

Work through the essentials#

Complete the checks for the modules you have configured.

1. Confirm the source information

   Open a product page, delivery policy, and returns page. Check that they are current, public, and included in the connected knowledge.

2. Test a useful answer

   Ask a specific product question and a policy question. Check the exact variant, any quoted price, and every destination link.

3. Test a limitation

   Ask about information your shop does not publish. The experience should explain the limitation or offer a useful next step instead of inventing a shop policy.

4. Test the storefront controls

   Open, close, and reopen the widget. Navigate to another page, test on a small screen, and confirm that consent and visibility behave as intended.

5. Review the first evidence

   After a real test conversation or advisor journey, check the relevant activity view. Allow processing time and distinguish preview activity from customer activity.

Agree on a support route#

Make it clear who reviews weak answers, updates source pages, and handles customer follow-up. Keep a working contact route available for questions the automated experience cannot resolve. If contact capture is enabled, verify its wording and the expected follow-up process.

After launch#

Review a few complete conversations and recommendation paths regularly. Improve repeated issues at their source, then repeat the same test. Record changes to prompts, product scope, or placement so a change in usage has context.