Recently I stressed the importance of contracts that define effective service-level agreements and include escalation procedures for ensuring effective enforcement. Many companies gloss over SLAs, or tack them on at the very end of the negotiation cycle. That’s a big mistake. SLAs should be as integral to the negotiation process as price, and telecom managers should craft them carefully.
I touched on the critical components that should go into an SLA. Now let’s talk about them in depth:
Performance. SLAs should specify end-to-end latency, jitter and packet loss. If the provider offers multiple QoS classes, the SLAs should be defined for each class. And SLAs should include average and maximum levels both for individual circuits and the network as a whole. Additionally, penalties should accrue both for chronic and acute failures to comply. In other words, there should be penalties for consistently failing to meet SLAs, even if the performance is only slightly worse than contracted for (regularly delivering 60 ms end-to-end delay instead of 50 ms, for example), as well as for extreme failures to comply such as a single instance of the latency spiking up to 200 ms. Naturally, this means that companies should plan to monitor and record performance on an ongoing basis – in addition to the monitoring statistics the service providers provide.
Provisioning. SLAs should specify average and maximum provisioning times for all types of services, including new-circuit installations, turning up VPN services, and activating calling cards and other voice services. Moreover, the excuse that “it’s all the local exchange carrier’s fault” is now increasingly out-of-date: If AT&T, for example, is now part of SBC, then the combined company is responsible for services from soup to nuts (at least within SBC’s region of operations). No more excuses.
Response and repair times. Service providers should specify exactly how long it will take them to respond to trouble-tickets and requests, and how long it will take them to restore service. And once again, telecom managers shouldn’t be content with averages or mean times – they should ask telcos to define maximum response and repair times. Because extremely rapid response-and-repair times may cost more, it’s reasonable for telco managers to define tiered response-and-repair times for sites with different availability requirements. This also can be helpful with internal chargeback: Telco managers can charge more for sites that need higher uptime.
Escalation procedures. Providing – and adhering to – detailed escalation procedures is the mark of a superior service provider. Telco managers should know exactly what will happen in the event that the above SLAs are breached, when the issue will be escalated and to whom. Carriers should proactively notify customers when an issue has been escalated and provide regular updates, even when the problem has yet to be resolved.
Getting all this into the contract isn’t always easy and often requires several rounds of negotiations. Telecom managers should be prepared to invest the time and energy upfront. They’ll be glad they did.
