The end-to-end activation SLA that no provider owns
Calder Telecom sells a managed 'provision and activate enterprise broadband' service to 1,200 business customers. Delivery crosses sales, three external providers (access network, core network, and installation), and internal field engineers. The customer contract promises 99.5% of orders activated within five days, with a service credit of 2% of monthly charges for misses. Each provider contract contains a 99.9% component availability SLA. Last quarter, end-to-end on-time activation was 91%, yet all three providers reported SLA attainment above 99.8%. Analysis shows 60% of the delay occurs at handoffs between providers, particularly when access network completion is not passed to installation. Providers reject an end-to-end penalty because they say they do not control the full chain. The sales director wants to renegotiate all contracts; the CFO wants to avoid a costly rewrite. The SIAM service integrator has been asked to propose a measurement and accountability model.
1. Explain why all providers can meet their component SLAs while end-to-end activation fails. Design a measurement architecture that would expose the real performance.
Read the worked answer
The providers can meet their SLAs because those SLAs measure component availability, not the end-to-end order flow. A provider's clock may stop when its own task is complete, excluding queue time, handoff acceptance and dependency waiting. Three 99.9% component measures do not compose into a 99.5% end-to-end promise, because the end-to-end service includes handoffs and shared dependencies that no single contract measures.
The measurement architecture should have four layers. First, an end-to-end SLA owned by the service integrator: percentage of orders activated within five days, measured from order acceptance to customer activation. Second, underpinning OLAs at each handoff: time from access-network completion to installation acceptance, and time from installation acceptance to activation. Third, process metrics inside each provider: queue time, first-time-right rate, rework and hold reasons. Fourth, experience measures: customer-reported activation clarity and repeat calls.
Crucially, the end-to-end clock must not pause at organizational boundaries. A shared order ID and event timestamps from all parties are needed. The 60% handoff delay then becomes visible as a measured OLA breach rather than an unattributed gap.
2. Allocate accountability for the end-to-end service and propose a contractual mechanism that avoids a full rewrite of all provider contracts.
Read the worked answer
Accountability should sit with the service integrator for the end-to-end outcome, while each provider remains accountable for its OLA and handoff obligations. The integrator owns the service chain map, the shared order ID, the end-to-end dashboard and the escalation of handoff failures. Providers own their task completion, data timeliness and acceptance of handoffs within the agreed OLA. Calder's sales and field engineering functions must also be in scope because they participate in the chain.
Contractually, a full rewrite is unnecessary. The existing contracts can be varied through a SIAM schedule or OLA annex that references the end-to-end service, defines the handoff events, sets OLA targets, and requires data exchange. Service credits can be back-to-back: the customer credit is funded by deductions at the OLA level where the delay is attributable, with a no-fault shared pool for genuinely shared delays. This preserves the existing commercial structure while adding the missing end-to-end obligation.
The mechanism must include an attribution rule. Without one, providers will argue about cause and the OLA will become another dispute. A joint attribution review, using timestamped events and a defined decision tree, is part of the schedule.
3. The first dashboard shows 99.6% internal availability but 91% customer on-time activation. How should the SIAM lead address the 'watermelon' reporting problem?
Read the worked answer
The gap is the classic watermelon pattern: green on the inside, red on the outside. The SIAM lead should not simply add the 91% figure to the existing report. The reporting model itself must change.
First, separate internal component reporting from end-to-end service reporting and make the end-to-end measure the primary service review metric. Component green is context, not the headline. Second, publish the service chain with the 60% handoff delay visible by provider and by handoff, so the discussion moves from 'our SLA is green' to 'this handoff is the constraint'. Third, define a single attribution method in advance and have it agreed by the providers, so the review does not become an evidence contest.
Governance should also change. The monthly review should start with the customer outcome and the handoff OLAs, then move to provider performance. Providers should present corrective actions for their OLA breaches, not defend their component SLA. The SIAM lead should escalate repeated misattribution to the SIAM board.
Finally, the lead should test whether the internal 99.6% is itself credible. If it excludes queue time or handoff waiting, it is not a performance measure; it is an availability measure being used for the wrong purpose.