Not all production support organizations have (or want) formal agreements that describe the level of service they will provide to their customers. Often, I think this can be a mistake.
If the service an organization provides is insufficient, service-level agreements (SLAs) and their performance metrics can help assure that the organization and its customers are measuring by the same yardstick and focus improvement efforts where they need to be.
But if your organization generally provides good service to your customers, SLAs and the metrics they encourage can be excellent communication tools and marketing materials for your stakeholders.
I was talking with a project manager for a county government with an extensive telecommunications network that provides radio service coverage to remote areas. An upgrade a few months ago led to some reliability issues that had been largely addressed, but her customers had become very sensitive to service disruptions, as customers will.
Her customers consist of other county departments, cities, and fire districts within her county. While she does not have a formal SLA with them, her organization’s historical performance had been exemplary. There are published benchmarks for reliability and availability, and the performance of her team over the past few years was significantly better than average in her sector.
The weekend before we spoke, a heavy storm had disrupted microwave communication and resulted in a “fail over” where the network corrected itself and routed traffic around a troubled link. Total duration of the service disruption was about five seconds. Her clients received a notification from the system of the reroute and, sensitive to any disruption, become concerned when they saw the outage on their consoles.
I asked her whether she had an SLA and if she regularly published or made metrics available to her customer base. She confessed that metrics might have been an embarrassment during the troubles a few months back.
What I suggested to her (and what I’m sharing with you) is that metrics can be a way of managing customer expectations. When things aren’t going well, publishing metrics lets clients know how badly things are going, which can be difficult—but it also gives you a way to show when things return to normal.
Operations and support is largely a thankless job. When you are doing your job well, no one notices. When you screw something up (or when something beyond your control interrupts service), customers who are used to your invisible competence can get testy quickly.
Publishing regular metrics can go a long way toward managing customer expectations and helping them appreciate the work you do. Better service is always possible if customers are willing to pay for it, but once good service is established, the cost of continued improvement can often be exorbitant.
Ultimately, establishing service-level agreements and tracking performance metrics such as the amount of system unavailability, the time to recover from outages, and the time to resolve errors of various severities can make your work visible and help customers appreciate your efforts.
