It is a common practice to measure incident resolution times and have SLA limits for different priorities. For example, a typical SLA can state that 90% of priority level 2 incidents must be resolved within 8 business hours. One complication is that sometimes resolution must wait for some information or third party act. Usually this waiting time is subtracted from the resolution time which means that the actual resolution time can be anything but be still within the SLA limit.
Another problem is the priority. It is hard to define the right priority and sometimes there is a need to change the priority as there is more information available concerning the nature of the incident. There was recently a discussion on the subject of changing priority where I commented that I consider the whole concept to be a bad ITIL practice. A few commentators disagreed. One person wrote:
… I very much disagree regarding ”bad practice”. Without a measurement for timing – and logically, the timing to resolve something more damaging would be shorter – one has only chaos and is at the mercy of whoever decides timing was not what it should have been.
In this article, I will explain the reasons why I consider it as a bad ITIL practice and what would be a better practice. The above comment is basically right; it makes sense to measure timing and it is true that important matters should be resolved faster. I disagree that it leads to chaos if customers can decide what is the right timing. But this is not the real problem, the core problem lies in the SLA connection. When resolution times are set as a SLA target the timing easily becomes dominant and it will override common sense and customer value.
Any metric can be harmful if it used incorrectly. Usually management wants to have numbers and easily defined targets but metrics can be toxic.
Road side speed measurements do not show high speeds, if they did, some drivers would try to get record numbers and the safety measure would become a source of danger.
Firstly, the measurements are very easy to manipulate. If you haven’t seen cases where all SLA’s are met but customers are quite dissatisfied, you do not know much about real life in ITSM. It is far too easy to play with the measurements as so many things are hard to define. Here are some techniques:
- ask difficult questions from the customer and stop the clock while they try to answer the questions
- give low priority to difficult cases, or change the priority if the deadline approaches
- classify more automatically generated events as incidents and solve them fast
All these tricks will help to fulfill the SLA promise without providing any value to the customer.
The second major problem is the setting of priorities. It is difficult and usually there are simple rules, which give a ticket a priority based on the affected service. The given priorities do not necessarily reflect the true business value related to the case. One thing is sure, any mechanical, automatic priority system will fail.
Here is an example, I’m sure many have seen similar cases:
IT service provider ITSP has a culture of fast responses and close cooperation with their customers. They know when they need to drop everything and jump to prevent a potential failure. They have processes and use a ticketing system to make sure that things are not forgotten, but they are not orthodox about it and do not always create tickets. They have no SLA’s.
One day ITSP management decides to implement best practices. All incidents need to be handled following SLA requirements and it is a severe error to let SLA times slip.
So, when next time there’s a potential failure, the IT staff concentrate on following orders and refrain from jumping to prevent the failure. The staff closes a group of minor tickets which are close to breaching the SLA limit before they start working on the major failure and resolve it just within the SLA.
Everything is ok, there has been no SLA breaches but the customer is mad because they could see that the IT people were closing insignificant tickets while their business ground to halt due to a major IT failure which was waiting.
The solution to the priority and SLA problem is simple; trust. If you can trust your service provider, you do not need to set SLA penalties. If you can trust your staff to make good decisions, you do not need rigid prioritization.
This far from easy, trust needs to be earned and it is easy to lose it. On the other hand, it is very rewarding and it is good for business.
Filed under: asiakaspalvelu, ITIL, Service Desk | Tagged: customer service, service desk, setting priority, SLA | Leave a comment »