For your team / Operations

Make the operating contract measurable before it goes live.

You are the person who has to run this on a Tuesday when the source system is down and the volume is double. Queues, clocks, failure states and the runbook get defined before launch rather than discovered in production.

Every queue needs an owner and a clock.

Volume, aging, failures, and fallbacks stay visible during ordinary and bad days.

  1. 01Population loaded
  2. 02Attempt state
  3. 03Aging clock
  4. 04Named owner
  5. 05Dependency failure
  6. 06Fallback and reconciliation

Operational completion never substitutes for clinical disposition.

Every queue needs an owner and a clock. Volume, aging, failures, and fallbacks stay visible during ordinary and bad days. The diagram shows Population loaded, Attempt state, Aging clock, Named owner, Dependency failure, Fallback and reconciliation. Operational completion never substitutes for clinical disposition.

Queue anatomy

Every stage records an owner and a timestamp

A queue you cannot inspect is a queue you cannot staff. This is the record a case produces, stage by stage.

StageWhat is recordedWho owns the next action
EligibleInclusion rule, eligibility date, population definition, source of the listAutomated against the agreed rule, auditable against your source system
AttemptedChannel, attempt number, timing, outcome of each attempt including failuresAI agent, inside the approved attempt pattern
ReachedDisclosure delivered, consent handled per program, intent capturedAI agent
Exception raisedException type, context, permitted actions, prohibited actions, destinationRouted to the owner the program names for that exception type
AcceptedAcceptance timestamp and accepting owner. Assignment alone is not acceptance.Navigator where included in scope, or your named owner
Action takenThe permitted action, the result, and any remaining dependencyWhoever accepted the work
Final stateCompleted, closed with a defined reason, or handed to an owner who accepted itThe accepting owner
ReconciledMatch or mismatch against your source of truth, and the reason for any mismatchJoint. Reconciliation failures are reported as their own line.

Escalation is not a final state. A case that was escalated and never accepted shows as unaccepted, aging, in the daily view.

Service levels and staffing

Defined for your program, not promised on a website

We will not publish universal response times, coverage hours or staffing ratios, because they are not universal. What we will do is write them into the program before launch and report against them.

The same applies to staffing. Whether navigator support is included, how many, which hours, which permissions and which systems they are provisioned into are engagement decisions. A page that implies a standing bench for every contract is telling you something that is not true.

  • Coverage hours and the behaviour outside them
  • Acceptance and first-action targets by exception type
  • Aging thresholds and the escalation path at each one
  • The named fallback owner and what activates them
  • Volume assumptions and what happens above them
  • Change process, release windows and rollback

Review one queue and one month of exceptions.

We will map the stages, the owners, the clocks and the failure states, and tell you which parts we would not take on.