In a service level agreement (SLA), a supplier agrees to achieve defined levels of performance and a customer obtains rights and remedies if the supplier fails to achieve those levels of performance. Whether the service level agreement articles are a stand-alone document or a part of a larger outsourcing agreement, it can be an effective tool for aligning the incentives of an outsourcing supplier with the objectives of an outsourcing customer.
The first and most important step in developing effective service level agreement articles is to ask the right questions. This article will give you those questions and some background on how to choose the right answers. The questions are as follows:
Service levels are defined ways of measuring a supplier’s performance. A service level generally is a measure of the quality, speed, availability, capacity, reliability, user-friendliness, timeliness, conformity, efficiency or effectiveness of services. For example, an availability service level for a computer system might be the percentage of the relevant time when the computer system is capable of performing a specific task.
A good service level is both within the supplier’s influence or control and an important measure of the success of the deal. A good service level is also designed to align the incentives of the supplier and the customer. For example, because a fixed price contract gives the supplier an incentive to cut costs (and quality) to increase margin, the SLA for a fixed price contract should focus on quality and timeliness.
The parties must define the service level with precision. For example, is a computer system “available” if its CPU is working, or do the databases and telecommunications systems also need to be working? Does it need to be “available” to the end user, who may not be able to access that computer system because of a local area network failure? Is it “available” when the operating system is working, even if the application program has failed?
Unless each service level is precisely defined, the parties will not have a true agreement — a meeting of the minds — on service levels. Instead, you’ll have a situation where the customer believes that the service level measures A, B and C and the supplier believes it measures only A and B. When C fails, you’ll have a dispute that the SLA doesn’t address.
For each service level, you need a process for measuring performance that the supplier achieves. For example, you could measure a computer system’s availability software continuously running within that computer system, periodic polling by another computer system, user complaints about downtime, or a monthly user satisfaction survey asking about perceptions of downtime.
The measurement process will affect the results. In our example, availability will be lower measured by a software system that picks up every millisecond of downtime than it will be if measured only by user complaints. Thus, you need to know how service levels will be measured before you can decide how high to set the bar for performance.
The key issues in choosing a measurement process include:
The measurement period is the time horizon for measuring performance. Typically, the measurement period will be a month or quarter. Longer measurement periods give the supplier more opportunity to make up for bad performance. Shorter measurement periods give the supplier a “fresh start” more often. Longer measurement periods mean that more is at stake during any one measurement period.
The service level agreement articles should define not only how long the measurement period is but also how much of that time is within the measurement period. Is it 24×7 or 9×5? Does it include times when the supplier is shut down due to severe weather, acts of war, terrorism, failure of its suppliers or other traditional force majeure events? Does it include periods when the demand for services exceeds the assumed levels?
The SLA should require the supplier to make available clear, useful and timely reports on performance for each measurement period. The SLA should define precisely what information will appear on the reports, such as exception reports for missed service levels and trend reports for key service levels. The service level agreement articles might also require the supplier to conduct a root-cause analysis of service level failures and report the results to the customer.
Note that we’re at question number six, not question number one. Too often, people start with the number — say, 99.9% — and then start to define a service level. That’s the wrong order because the number is only meaningful when you know what’s being measured, who will measure it, how it will be measured and how often it will be reported.
The SLA can include both minimum service levels and target service levels. The supplier would be obligated to meet any minimum service levels, and failing to do so would be a breach of the outsourcing contract. The target service levels would be measured, and there might be service level credits associated with failure to meet them, but a failure generally would not be a breach of contract.
Customers should be wary of these pitfalls:
The SLA may include not fixed but floating performance commitments. Particularly in long-term, large-scale outsourcing agreements, service level agreements may increase the performance commitments through:
A “service level credit” is a credit that the supplier grants to the customer after a service level failure. The supplier may be required to write a check to the customer or the customer may simply have the right to apply the credit to future service. Either way, it reduces the effective price of the services and the supplier’s profit margin.
As an example, an SLA might call for service level credits for any of 10 service levels. For each of those 10 service levels, the SLA might indicate a number of “credits” to be granted upon a failure, with each “credit” being a small percentage of the customer’s total bill for the measurement period.
The total service level credits for a measurement period might be capped at, say, 10% or 15% of the total monthly bill. This means that, although poor performance could reduce the supplier’s profit margin, it would not create a loss for the supplier. The total pool of service level credits (that is, the total service level credits payable if the supplier misses every service level) would then be some multiple of the cap.
There are many variations on this theme. Some SLAs impose credits only for repeatedly missing required service levels. Some SLAs allow the supplier to earn back service level credits for subsequent good performance, or to use superior performance to get “Get Out of Jail Free” cards to avoid future service level credits. Some SLAs have entire schedules of credits for a single service level, with larger service level credits for larger or repeated violations.
Service level credits are an incentive system. Smart customers design the service level credit structure on obtaining good performance throughout the contract term. They also retain the right to revise the service level credit structure so that they can re-align the incentives as their priorities change.
One important question is whether the service level credits are the customer’s sole remedy for a breach or merely one of the customer’s remedies. This is an area where loose contract language can have surprising results.
A service level bonus is an increase in the price triggered by performance above the target service levels. They are far less common than service level credits because, generally, the customer buys the level of service that the customer needs and is not inclined to pay for better service that it needs. Generally, if the supplier has an opportunity for increased compensation, it is tied directly to creating business value through, for example, a commission on sales made through the supplier’s call center or a gain sharing arrangement based on actual measurable gain for the customer.
Outsourcing agreements, generally, can be terminated for cause upon a material breach. A sufficiently severe service level failure would be a material breach. However, without clear language in the SLA, the parties might argue about whether a service level failure is sufficiently severe to be a material breach.
The SLA can provide increased certainty by defining particular events that, without argument, allow termination for cause. You can set minimum service levels at the level that allows termination. You can also define an amount of accumulated service level failures that allow termination for cause. For example, termination for cause might be allowed if the supplier breaches a single service level three times in succession, or the supplier breaches enough service levels that the service level credits are limited by the cap. The effect is to give the customer a clear exit right for substandard performance.
Attorney Brad Peterson is a partner in the IT and Outsourcing Practice at Mayer, Brown & Platt in Chicago. He is the co-author of The Smart Way to Buy Information Technology: How to Maximize Value and Avoid Costly Pitfalls (AMACOM Books, 1998). You can reach him at [email protected].