Improvement Targets in SLA's

[Mary-anne]

I'm looking for ideas around writing in Improvement Targets into SLA's. This would around the concept of continuous improvement over time where you would have modify the service levels on each anniversary of the SLA to reflect this concept.

I have seen many lengthy formulas used in the past and am struggling to understand the reasoning behind the formulas.

As one would expect to see continuous improvement over a partnership of many years, and I would want to document this in a SLA without having to base it on something that seems to have no rhyme or reason, does anyone have any other suggestions on how this cold be better handled and documented?

[rjp]

Hi,

Are you talking about improving the Services themselves, new capabilities, higher production throughputs etc., or improving on how well you are delivering the Services as they are defined - ie. fewer SLA breaches, fewer incidents, etc. (or maybe both? - they are never completely distinct)

In general I would be cautious about writing improvement targets into SLAs. The primary function of an SLA is measuring actual delivery against required (and agreed) delivery. And automatically racheting up Service levels each review cycle is not necesarily a good idea.

Remember the highest level in the IT organisational maturity model is where ICT is driven by it value to the business. Which is to say the business partners with whom SLAs are negotiated should be driving the Service 'level' targets according the the value they return to their business activities. Not a structured in 'principle' of automatically expected increments. And while Moore's law seems to still have legs, it's effectiveness in a business situation is attenuated somewhat by the law of diminishing marginal returns.

In some cases ramping up Service levels may serve to generate no additional value for the business - though it might make IT look good provided costs aren't being ramped up correspondingly. There will even be times when alignment with business requirements means some services should be scaled back.

A service improvement plan should be a separate animal to your SLAs. It would be a more strategic document, one of the objectives of which would be to express the overall operational plan for the IT organisation in terms of its ability to support the underlying ICT requirements of the current business plan.

Both the IT operational plan, and the service improvement plan (which may be a component of the former) would be a reference document when reviewing and negotiating SLAs with your customers. The SLAs operational targets would account for the planned improvements to services - it would also (critically) flag areas where for sound business reasons specific capabilities might be being wound down.

Spoons for the soup, forks for the salad - IMHO

[rjp]

..addendum...

yes, I meant to say a little something about the 'formulae' - of which I have seen a few I also thought to be of dubious worth.

The term 'level' is a bit slippery. I accept the need for objective measurements - but not all services are easily quantifiable to numbers on a 'dashboard' graphic. I seriously doubt whether the fetish with email availability percentages expresses the real delivered value.

Which is to say 'values' is the real goal - not an statistical 'level' - and value is harder to quantify, and frequently includes a large number of closely related variables. For example better training programs for end users of Exchange/Outlook may have a far greater impact on the value of the service to the businesses productivity than increasing uptimes from %99 to %99.5.

Service delivery like you said is a partnership....

[rjp]

..addendum...

yes, I meant to say a little something about the 'formulae' - of which I have seen a few I also thought to be of dubious worth.

The term 'level' is a bit slippery. I accept the need for objective measurements - but not all services are easily quantifiable to numbers on a 'dashboard' graphic. I seriously doubt whether the fetish with email availability percentages expresses the real delivered value.

Which is to say 'values' is the real goal - not an statistical 'level' - and value is harder to quantify, and frequently includes a large number of closely related variables. For example better training programs for end users of Exchange/Outlook may have a far greater impact on the value of the service to the businesses productivity than increasing uptimes from %99 to %99.5.

Service delivery like you said is a partnership....

[Mary-anne]

Thanks for the reply rjp, your comments about a seperate Service Improvement Plan have been well considered.