What behaviour do your SLA targets encourage?

2015 01 30 What behaviour do your SLA targets encourage IMAGEService level agreements (SLAs) should help to improve services, by making sure that both the service provider and their customer are very clear about responsibilities and expectations. If you don’t agree targets with your customers then it can be very difficult to plan and manage services, and if customer expectations are not explicit they can be very difficult to meet.

Unfortunately, many SLAs have very poorly thought out targets, and these can often result in the service provider behaving in ways that are the exact opposite of what the customer wants. This is especially true if SLA targets are tied to rewards or penalties.

We all know that SLA targets should be SMART:

  • Specific: The target is clear and unambiguous
  • Measurable: You can quantify or track the value of the metric for this target
  • Achievable: It is possible for the goal to be reliably and regularly achieved
  • Relevant: Achieving the target is relevant to the mission or objective of the organization
  • Time based: The target includes the timescale over which it should be measured and reported

But this is not enough, because even SMART targets can encourage the wrong behaviour, and result in worse outcomes than no targets at all.

I was reminded of this issue when a friend told me about a problem with a contract for outsourced security vetting. This contract included a requirement for 95% of assessments to be completed within 4 days. The outsourcer was consistently achieving this by not making any effort at all to carry out screening for difficult cases, using all their resources to carry out simple screenings. This achieved the required metric, but the customer was very unhappy as they had many people waiting indefinitely for screening that made no progress at all.

An example that I saw myself, in an outsourced IT environment, was just as bad. The SLA included the following terms:

  • 99% of priority 1 (P1) incidents to be closed within 4 hours
  • 95% of priority 2 (P2) incidents to be closed within 8 hours

Each of these metrics was to be reported monthly, based on the previous 3 months data, for example the 1st July report was based on incidents that occurred in April, May and June. There were significant financial penalties for failing to meet the targets. One month the service provider knew that they would not meet the target for priority 1 incidents, so they sent an email to their support personnel which said “The P1 target will be missed this month, but we can still meet the P2 target if we focus all our resources on P2 incidents, so make sure that you deal with any P2 incidents before you work on P1s”. When the customer found out about this instruction they were incredibly angry. The service provider had effectively told their staff to focus on preventing financial penalties, rather than on meeting the customer’s need to have higher priority issues resolved first.

A final example, just to show how pervasive the problem is, was reported recently in the Guardian newspaper. Publication of patient death rates for heart surgery has resulted in “surgeons being less willing to operate on patients with a perceived high risk of mortality, despite potential overall patient benefit”.

So what can you do to avoid these traps, and write SLAs that will do the job they are supposed to?

One option is to avoid including penalties and rewards in SLAs. I have never seen an SLA of this type that really led to increased customer satisfaction or service provider performance compared to an SLA that did not include penalties and rewards. It seems to me that service providers will always find a way to make the numbers come out right, even when the service is terrible!

Another thing that you should do is to evaluate every proposed target by asking yourself “What behaviour might this target drive?” or “What could the service provider do to try and achieve this target, even if they aren’t delivering the service that I want?”  To help remember this you could use the acronym SMARTD when evaluating targets, every target should be Specific, Measurable, Achievable, Relevant, Time-Based and

  • Driver: The target encourages the behaviour you want, and discourages behaviours that you don’t want

How recently did you evaluate the targets in your SLAs? Are you sure that they all drive the behaviours you want?

 

Image credit Pink Sherbet Photography

comments powered by Disqus

Optimal Service Management Ltd.

7 Ingatestone Road, Woodford Green,
Essex, IG8 9AN, UK

Registered No: 8791379 England

Phone: +44 20 8504 2002

Recent Posts

  • 2022 02 15 Risk appetite
    Defining your risk appetite

    How to create simple definitions of risk appetite levels, and then assign these to each of your organization’s projects, services, business units or any other clearly identifiable part of your work.

  • 2021 11 25 Mentoring inage
    Mentoring 101

    Mentoring is a great way to develop both professionally and personally, and the mentor can gain as much from the relationship as the mentee. This blog gives an overview of how you can get started as a mentor, or as a mentee.

  • 2019 09 11 A great customer journey has to be planned from end to end
    A great customer journey has to be planned from end-to-end

    Have you tried mapping out your customers’ journeys? If not, then it’s an exercise well worth doing.

Latest Tweets