The SLA Nobody Can Meet
A managed-service provider has an SLA promising 99.9% availability for a client's order-management system and a 4-hour resolution target for Priority 1 incidents. Last quarter the system delivered 99.4% availability across three unplanned outages (2h, 5h, 6h). The client's business runs 24/7, but the provider's support is 08:00-18:00 weekdays with best-effort out-of-hours cover. The contract was signed two years ago when the system was lower-volume. The provider's costs to meet the 4-hour target out of hours are rising, and two engineers have flagged burnout. The client is invoking service credits. The account director asks whether to renegotiate or invest to meet the SLA.
1. Assess the gap between what was promised and what is being delivered, and identify the root causes.
Read the worked answer
Promised vs delivered: 99.9% allows about 43 minutes of downtime per month (~8.7 hours/year). Actual was 99.4%, roughly 13 hours of unplanned downtime in the quarter, far outside the commitment. The 4-hour P1 target is also structurally at risk because support hours do not cover a 24/7 business, so a P1 raised at 19:00 may wait until morning.
Root causes:
- Contract designed for a smaller, less critical service; demand and criticality grew.
- Coverage model mismatched to a 24/7 requirement, guaranteeing breaches and burnout.
- No evidence yet of whether outages are repeat failures (problem management gap) or one-off; the 2h/5h/6h pattern deserves root-cause review.
The gap is a service-design and contract-alignment failure, not simply engineer effort.
2. Recommend a course of action and justify it against the alternatives.
Read the worked answer
Recommend a structured renegotiation backed by a short investment to stabilise, rather than promising to 'just meet' the SLA.
Why not invest to meet as-is: 24/7 four-hour cover needs a rota the current cost base cannot sustain, burnout is already visible, and the client's service credits signal the relationship is at risk. Buying your way to 99.9% without fixing the coverage model repeats the mismatch.
Why not simply renegotiate downward: it looks like retreat and may breach the client's own commitments.
Recommended path:
- Present the availability data honestly and agree a realistic target (e.g. tiered SLA by criticality, with defined out-of-hours cover).
- Fix the availability causes via problem management before re-committing.
- Offer a transitional credit or service improvement plan to rebuild trust.
This aligns what is promised with what can be delivered and funded.
3. What would you put in the service improvement plan and how would you measure it?
Read the worked answer
Improvement plan contents:
- Problem management on the three outages to find and remove repeat causes.
- A defined out-of-hours support model (rota, escalation, on-call) sized to P1 needs, with a costed option for the client.
- A revised SLA: availability target achievable and monitored, P1 response/resolution differentiated by time of day, and clear priority definitions.
- A joint review cadence with the client.
Measurement:
- Monthly availability against the new target, with trend.
- P1 response and resolution times, split in-hours vs out-of-hours.
- Number of repeat outages (target: zero repeat causes).
- Engineer on-call load and attrition as a sustainability guardrail.
- Client satisfaction and credit exposure as the commercial signal.
Report against baseline so the client sees improvement, not just a lower bar.