Write down the service you intend to deliver

Example: employee laptop and desk setup
DecisionExample policyCommon mistake to avoid
OutcomeLaptop and desk setup ready for the employeeDo not call the request complete when only a purchase order is approved.
Required contextBusiness need, location, date, and equipment requirementsCollect facts that change the delivery decision, not every possible asset field.
AuthorizationDesignated equipment reviewerA review decision permits the commitment; it does not prove physical delivery.
Delivery tasksPrepare laptop; deliver dockEach task needs an owner and completion details.
AcceptanceEmployee confirms both deliverablesAn incomplete or blocked task prevents final acceptance.

Build the process around its decisions

  1. Separate requests from unexpected problems

    “Provide a laptop for a role change” is a defined service. “My existing laptop will not charge” is a support issue. The people may overlap, but the input, decisions, and end conditions are different.

  2. Name the owner of the outcome

    Assign a service owner who can resolve ambiguity across teams. Task owners do their parts; the service owner remains accountable for the complete request and the promised outcome.

  3. Require authorization only where it changes the decision

    Spending, access, or an exception may need a human reviewer. Routine low-risk delivery may not. Avoid an approval ceremony that delays work without adding a meaningful control.

  4. Make incomplete information recoverable

    A reviewer needs a way to return a submission, explain the missing detail, and receive a corrected version. Do not erase the original decision or turn a rejection into a silent field edit.

  5. Define what counts as delivered

    List the required tasks and the evidence each owner records. The employee should know what they are accepting. One completed component must not hide an unfinished dependency.

Different requests need different handling

Business-system access

Collect the system, requested role, business reason, and expiry date where needed. A reviewer authorizes access; an authorized administrator grants it in the actual system; the result is recorded for the requester. A status change in the request app is not account provisioning.

A broken meeting-room display

This is an unexpected issue rather than a service-catalog purchase. Capture the room and impact, assign a handler, record diagnosis and waiting, and confirm the display works again. Use the internal help desk example for that lifecycle.

Roll out one complete service first

  1. Begin with one service and a few realistic cases

    Use a routine request, a request needing authorization, a partial delivery, a returned submission, and a completed delivery. Ask the actual service owner whether each case leads to the correct next action.

  2. Review stalled work before adding dashboards

    Look at missing information, unowned tasks, blockers, and acceptance delays. Build the view that helps a person resolve those conditions; a count alone does not explain what to do.

  3. Expand with deliberate boundaries

    Add another service when its inputs and delivery can be described. Keep confidential requests, technical provisioning, and specialist incident response in the systems and permissions designed for them.

Planning a service request process

Who should own service request management?

Assign a service owner who understands the promised outcome and can resolve cross-team handoffs. Individual delivery tasks still need their own responsible people; ownership of a dashboard is not ownership of the service.

Should we design the policy before choosing software?

Define at least the service outcome, required information, authorizer where needed, task owners, and acceptance condition first. Then check whether the app makes those decisions and exceptions usable.

What should happen to an incomplete request?

Return it with a specific question and preserve its history. Make it clear who must correct the information and when it can proceed; do not mark it rejected solely to get it out of an active queue.

What metric should a small team start with?

Start with a decision: which requests have no owner, which tasks are blocked, or which deliveries await confirmation? Add delivery-time metrics only after the start, finish, working calendar, and waiting policy are defined.

Are the service-delivery and help-desk examples one synchronized app?

No. They are separate examples for defined service delivery and unexpected employee issues. If you need a handoff between them, design the relationship and ownership rather than assuming automatic synchronization.

Turn one written service policy into a working process

Use the sample request and delivery tasks to test your own handoffs before expanding to more services.

Explore the service request app