Which brings me to today's subject, the service level agreement, or SLA. My theory is that SLAs are evil and should be dispatched back to the bureaucratic hell from which they were spawned. Okay, that's a little harsh. However, what I observe is that SLAs are divisive, and introduce an unnecessary element of contention into the relationship between customer service provider and customer. And I think that's especially true when it's an internal customer service organization, like an IT help desk.
So why is this? The first issue is that SLAs are usually developed entirely within the customer service organization without any input from the customer. Think about it, did your internet service provider consult you when they decided on their SLA? Did they include you in the decision to create four hour windows when they might show up to provide whatever service they're providing. I think not. Now the reason customer service providers do this sort of makes sense. After all, they know their resources, and they know how long it takes to perform a particular customer service function, so it's a simple mathematical calculation. And really well run customer service organizations adjust the SLAs based on feedback from (usually irate) customers, or lessons from other organizations. For example, the cable TV companies used to routinely have a full day window when the cable guy might show up, replaced by the four hour window that we typically now see. And some companies even have a two-hour window!
We also exclude customers from the development of the SLA because we're concerned that they'll ask for something unrealistic like immediate help. Well of course they will! After all, who doesn't want an IT person at their beck and call? But having a conversation helps us understand when that immediacy is critical and when it isn't, and it helps the customer understand our constraints. So if you must have an SLA, at least include the customer in the development or review of the agreement.
We also exclude customers from the development of the SLA because we're concerned that they'll ask for something unrealistic like immediate help. Well of course they will! After all, who doesn't want an IT person at their beck and call? But having a conversation helps us understand when that immediacy is critical and when it isn't, and it helps the customer understand our constraints. So if you must have an SLA, at least include the customer in the development or review of the agreement.
The other issue with the the development of the SLA is that customer service organizations want to develop SLAs that they can meet a high percentage of the time. So let's think about how the conversation might go when developing an SLA for something as simple as a desktop support request. "Okay, so the user emails a request, and it might take use 4 hours to get to the request because of our queue, then it might be a more specialized issue, so we might need Joe, and he's super busy, so we need another four hours, oh and what if he's on vacation, and then we might need to order a part from a vendor, and that takes overnight, and then it has to get through our Receiving department and then we have to schedule time with the client to change out the part. 5 days, that's our SLA, we can make that almost all the time". The thing is, we're developing the SLA for a worst case scenario that doesn't (or shouldn't) happen often. So most service requests actually are completed in a much shorter amount of time, because they don't need Joe, or a part. But the customer hears that you have a 5-day SLA, and what they interpret is that when you call that team, they'll get back to you in 5 days. And frankly, that just opens your team up to ridicule.
My final issue with the SLA is that it is used in a defensive way by service providers. I was recently in a fairly contentious meeting between two groups on my campus. The "customer" in this case has a well-earned reputation for letting things get down to the wire, and this situation was no exception. They were handing something off to a service provider, and it was really at the last minute, actually, beyond the last minute. The service provider reminded them that they had a 5 day SLA, and basically said, if you don't get it in by X, you won't get it until Y. What the customer heard was, I'm not even going to try to do this in a way to accommodate your needs. I decided I have 5 days to do this and I'm going to take every minute of those 5 days. In fact, I'm fairly confident that the service provider more often than not beats the 5 day SLA significantly. Even using different language would have helped in that case.
I was recently in a meeting with a potential vendor and we asked about customer service. The presenter never once mentioned an SLA. What he said was, we resolve 1/3 of our calls within an hour, 1/3 within a day and 1/3 take longer, because they're more complex. I think that this is a more sensible approach. So rather than providing an SLA, measure and publish your stats. It seems more reasonable to the customer. And it gives your team something more challenging, while still realistic, to aspire to. So have an SLA if you must, but don't wave it in front of your customer. Instead share your stats with them. It's more realistic, and probably looks better too. If it doesn't, well, then you have a different issue to address.