Trace one permitted test record from creation to storage and back into the interface. A realistic dashboard, a successful response or data surviving a reload does not settle the question. Look for the actual read and write operations, the environment they use and a matching stored record. Browser-only storage can be legitimate, but it should match what the app promises.
1. Define what “real” means for this feature
A static page can work correctly without a database. A personal calculator might intentionally save preferences only on one device. A shared customer portal makes a different promise: authorised users need access to the intended persistent records.
Write the expected result first. For example: “An enquiry created by this test user should appear in the staff account and remain available after signing in on another device.” This is an illustrative requirement, not a test result.
2. Inspect the read and write path
Ask the developer to identify where the displayed list comes from and what happens when you add or edit a record. Chrome's Network documentation explains how to inspect browser requests and responses.
Check the relevant endpoint, response content and deployment environment. A real API can still return sample fixtures. Conversely, an absent Fetch/XHR request does not prove the absence of a backend: server-rendered pages, persistent connections and caching can change what you see. Use the implementation and service records to resolve uncertainty.
3. Use a unique, harmless test record
In your own authorised test environment, create an item with an unmistakable label. Locate the matching record in the intended service or database through the normal authorised tools. Check its identifier and the values you entered.
Then read it from a separate browser profile or another device, using an account authorised to see that record. If the product is intended to share data, this helps demonstrate that the result is not confined to the first browser profile. Compare it with the matching record in the intended backend. Keep the account permissions and expected ownership the same as the real workflow. Do not use genuine customer information for this check.
4. Understand what a reload proves
MDN documents that localStorage can retain data across browser sessions. A record that survives refresh may therefore be stored locally.
Avoid clearing storage on a working account simply to investigate: you could remove useful local data. Inspect storage with the developer or use an isolated test profile instead. Local storage and remote storage may coexist, for example when drafts or cached records are involved.
5. Review mixed states and honest labels
Ask what appears when the service has no records or cannot be reached. Demonstration entries should not silently become apparent customer activity. Check whether the deployment has a deliberate demo mode and whether the interface explains it.
Matching an actual stored record establishes one part of the workflow. It does not prove that all data is accurate, permissions are correct or recovery works. Continue with the AI website readiness review before relying on the app for customers.
Sources
Reviewed 13 November 2026.