Step 4. Preparing Service-Level Agreements:
Effective service level agreements (SLAs) identify the expected results and the measures by which both
parties will evaluate performance. As binding agreements with the force of standard contracts, SLAs for
outsourcing frequently require sophisticated legal and technical expertise, because of their complexity
and variety. Importantly, most outsourcing failures occur because neither side has articulated how to
assess the progress made against a specific goal or target. Even when targets are clear, measures may
not be effective. Finally, the absence of a balanced set of performance penalties and incentives can
undermine the delivery of results, especially when expectations are not in alignment.
(One expert commentator recently suggested that clients intuitively expect service levels
in the top 10 percent, while providers consider the top 25 percent acceptable.) Accordingly,
clear performance areas and effective measurement are essential to outsourcing success.
In brief, a well-designed outsourcing SLA be made by describing standard requirements such as
start and end dates for the service, the schedule for reviewing performance, and documentation
to be used in measuring the service.
The agreement then should carefully delineate the scope of services, including:
- Assets, equipment and infrastructure to be serviced
- Types of services to be provided
- Decisions to be made by the provider
- Decisions to be made by the client.
Equally important is a breakdown of fees, with both penalty and incentive provisions clearly outlined:
- Amounts period by period
- Volume of work covered
- Quality of work to be provided
- Provision for over- and under-performance.
Good performance metrics measure results, not just activity, and they are kept to a minimum.
We have developed several guidelines for selecting outsourcing metrics, including:
- Aligning metrics with the business goals of the service.
- Selecting metrics that enhance the ability to diagnose problems, escalate attention, and remedy
- Avoiding metrics that distract from the service itself.
- Limiting the metrics for each service to one or two.
The following is a sample of the types of measures applied to data processing outsourcing activities:
- Availability, typically two clock times between which the system must be up.
- Hours required to fix a system error or interruption.
- Response time for a transaction, typically in seconds.
- Turnaround time for a batch job, typically in minutes or hours.
- Batch window, typically a clock time by which the batch must finish
- Network uptime and availability, including time to back up a failed line.
In outsourcing as in other service activities, the tightest service definitions produce the most