Example: a repair-request portal An illustrative brief, not a client result or a working portal. First release: submit a request, let a coordinator review it, then assign it to a technician. Payment, live stock, a mobile app and AI are not automatically part of this example. Three roles, different access A customer sees only their own requests. A coordinator reviews and assigns permitted requests. A technician sees only assigned work. Access is checked on the server; hiding a button is not an access-control test. A repeated click or a failed message Repeating submission must not silently create duplicate work. If the email service fails, keep the request visible to the coordinator and provide an agreed way to check and retry. A change after approval Adding payments or access to all historical customer records changes the scope. Assess the data, risks, cost and acceptance checks before changing the implementation. A project you can review in stages Each stage has a deliverable, price and acceptance decision. You know what is being approved before the next stage begins; a prototype is not mistaken for a production service. 1. Define one useful result Describe today's process, people, data and constraints. Separate the first release from later ideas. Record assumptions and anything that needs a feasibility check. 2. Walk through a prototype Review representative screens and their error or empty states using fictional data. Agree the scope, permissions, technical choices and checks before implementation. 3. Build and review the first version Implement the agreed journey in a test environment. Show progress and check roles, integrations, failed requests and repeated actions, not just the happy path. 4. Approve launch and handover Review the agreed checks, recovery route, access, costs and instructions. Launch only after approval; list any remaining issues and the support arrangements. What must be checked before handover Useful journeys and permissions Try the agreed roles and workflows, including attempts to read another customer's data. Verify server-side decisions and appropriate approval steps in the test environment. Connections and failure handling Test successful, failed and repeated operations with approved test accounts. Record what the provider confirmed and what remains unverified. Screens and practical use Check mobile layouts, keyboard navigation, forms, useful errors and every ordered language. Review load or accessibility tests only against a written, agreed target. Ownership and recovery Hand over the agreed source or configuration, accounts, deployment notes, dependencies, recurring costs and instructions. Check the agreed backup/export and restoration procedure and record support responsibilities.