>> continued from previous page
SLAs
Detailing every clause that your contract should contain would be tedious. You're likely to see a long preamble that forms a basis for agreement. That is usually followed by any number of sections, such as:
- Definitions
- Financial arrangements
- Schedules
- Services/products and SLAs (service-level agreements)
- Responsibilities of the parties
- Limitations of liability
- Termination conditions
Why so many sections? The contract should document every detail of your relationship with the service provider. It should include all the services the provider is responsible for and all the fees that apply to those services. (The adage "To err is human, to forgive divine" does not apply to contract law.) If it's not in the contract, there is no duty or obligation to perform on the one hand and no promise of payment on the other. Where there is a duty to perform in the contract, it should be accompanied by an SLA. To meet needs unforeseen at contract time, meeting and discussing changed circumstances on a regular basis can lead to modifications and addenda.
An SLA sets parameters for performance under the contract and sets a price for that performance. SLAs can apply to any service that can be measured objectively at a negotiated price. They serve to align enterprise objectives and supplier incentives to perform. They can also ensure that performance doesn't slip. Enterprises can obtain rights and remedies in case the supplier fails to perform at certain levels. Each SLA should define the supplier's service and the measure used to determine the quality of that service or performance. It should also define a period for measurement and any output reports used to monitor performance.
Give serious thought to what you measure. Done right, an SLA can measure the availability, reliability, timeliness and efficiency of a service. Done wrong, it can measure nothing. For example, end-user surveys will not provide quantitative details on hardware repair. They may, however, put a dent in your budget and add to your bedtime reading.
Good service levels fall within the supplier's control and correspond to both parties' business incentives. If you outsource service and support, proper metrics may go to the supplier's response time to an incident or the entire time it takes to fix an end system. These items can translate to the lost volume of work for unavailable computer resources.
The provider must establish specific performance levels and offer ways to measure those levels. The terms should be spelled out in sufficient detail to easily determine whether the supplier fell short of the mark, achieved it or exceeded it. For example, you can define resource availability by whether the CPU is powered or the system is available to end users. You could also define it by the number of supported connections or the transactions per second. If you don't define it with sufficient detail, there is no "meeting of the minds" on that particular SLA.
Once you've established the terms, the SLA should detail the process and the output results. Each party should have immediate access to the output. Are you going to measure by end-user complaints registered when a system is unavailable? Or will you use an enterprise management tool to detail network or system availability by the millisecond? Will the tool report on availability 24x7, 12x7 or 9x5? Using a more detailed and accurate process leaves less margin of error. This puts the supplier at a greater risk of nonperformance but results in higher system availability. It can also lead to higher costs for higher levels of performance. Other processes could bring in third parties. For example, the parties may agree on a standard metric for Web server performance provided by Keynote Systems, then split the cost.
Regulating the Relationship
Once you agree on the type and level of detail for reporting requirements, you can link the results to an appropriate regulatory framework to give your contract some teeth.
Failure to satisfy an SLA fixed at a certain level of performance can trigger a penalty clause. Penalty clauses vary by design, but keep in mind that these measures are regulatory, not punitive. They apply to a strategic partner, not an adversary. Penalty clauses usually define service-level credits that can offset monthly service charges to reflect minor faults in service levels. For example, if an SLA stipulates that PCs should be repaired or replaced within one business day, service-level credits for free repairs can accrue for substandard performance. Also, the cumulative time for unavailable computer resources could roll into a credit applied to a monthly service charge. For major faults--that is, long periods of unavailable services--you may want to increase the penalty measure. Another option may be to determine which party should bear the cost of insurance. Be clear on the terminology used for penalty clauses and make sure the metric is within the provider's control. Otherwise you will most likely suffer intermittent contract disputes.
Although penalties and credits are crucial to a strong contract, they reduce the supplier's profit margin while increasing the perceived risks. As a result, you may pay higher overall prices. To reduce the supplier's risk of failure, you can cap credit levels at some percentage of a monthly bill, or agree to accept credits for repeated failures only. If you don't do this, the supplier may deem the risk too high and walk from the negotiating table.
You can also offset some of the risks associated with penalties and credits by considering service-level debits or bonuses for performance that exceeds target levels. Such incentives are not common in SLAs; enterprises aim for properly scoped levels of service. But you may want to provide suppliers a way to generate business value based on their best practices and expertise. For example, if the enterprise outsources a portal service, the supplier could receive a commission on the number of sales or service requests that passed through the portal.
You can get creative with credits, debits and bonuses. Make sure they are detailed in clear, unambiguous language. And apply them to SLAs that impact your enterprise's bottom line, or your relationship will get mired in credit-debit disputes and you'll start to feel like a coupon king. You need not worry that these are your only remedies for service-level failures. Outsourcing contracts are like other contracts. They can be terminated based on a material breach of terms. These terms should be defined in an exit clause.
Every outsourcing contract should have an exit clause--some scenario that defines the end of the relationship. It can be as simple as the termination date of the contract or a notice to end the relationship when one party buys its way out. Buyout provisions for the provider and the enterprise can limit the total price to a certain percentage of the entire contract. SLAs can also define particular events that will terminate the contract for cause. Repeated failure to meet minimum service levels could trigger a termination procedure, for instance. Defining the grounds for termination in an SLA is a good practice, and the details should be worked out with the supplier. It can mean a smooth transfer to another service provider or the ability to transition the services back in-house.
Service-provider contracts should enable reliable services at standard levels of performance. If you consider your service provider a strategic partner and negotiate price and performance that manage the risks of both parties, you're dealing from both sides of the table and will most likely come out with a winning hand.
Sean Doherty is a technology editor and lawyer based at our Syracuse University Real-World Labs®. A former project manager and IT engineer at Syracuse University, he helped develop centrally supported applications and storage systems. Send your comments on this article to him at sdoherty@nwc.com.