Service Level Agreements – Making Them Work

How can the implementation of service levels be improved in small to medium enterprises?

A recent survey (June 2002) carried out by IT week in the United Kingdom showed that of more than 1000 companies contacted, nearly 50% were very unlikely set up an internal Service Level Agreement SLA for provision of IT services.

One of the reasons for continued reluctance of IT managers – and their staff – to buy into the idea of implementing clear SLAs is that it is still seen as a means of spying on the productivity of IT support staff and a means of restricting the support internal customers received.

If implemented half-heartedly, or with the minimum amount of preparation, this can be the case. When handled properly, and with the right approach, an SLA can provide far more benefits to all concerned, whether they are the internal customers, the support staff, or the organisation overall.

A good SLA acts, to a large part, to help manage user expectations but User Expectation Agreement, or “UEA” is a more unwieldy term by far.


A good SLA is bi-directional: it covers the expectations of both parties in the support relationship. From the support side of the equation, time limits for work to be carried out should be clearly set out. From the customer side of the equation, minimum notice periods need to be specified.


The crucial factor in a working SLA is that of universality: if an SLA is to work, it needs to cover everyone from the office junior through to the Board of Directors.

In most organisations there will naturally need to be a system of hierarchy priorities: the technical problems of the CEO may need to be resolved more rapidly than those of a receptionist.

It should – ideally – be the business impact of the problem that determines the priority of the issue, not the political impact. Context should be everything.

The Chief Financial Officer who is reduced to only printing his email in black and white, for instance, is of less immediate significance than the salesman who needs to produce a colour print out for a sales presentation.


No matter how thoroughly formulated and negotiated the terms of reference for delivery of support are there will always, always be occasional situations that arise requiring a step outside the rules in a critical situation.

This may be due to unforeseen circumstances, exceptional events or activities, or a change in the operations or environment that is being supported, whether temporarily or more permanently.

Documented, consistent escape clauses must be included so that breaking out of the terms of reference is handled consistently, and in approved circumstances. An effective, accessible approval chain needs to be in place for such situations so that exceptions to the terms of reference can be escalated quickly.

The periodic review of the terms of reference should include a review of the instances in which the escape clauses are called into effect, not only since the last review, but cumulatively since the last major revisions to the terms of reference.


Like all objectives in business – whether they are project milestones, annual performance appraisal targets – clauses in an SLA need to be quantifiable. If the factors cannot objectively be measured, they cannot easily be analysed. A numerically measurable quantity can be targeted for future improvement, and future monitoring.


Again, like all objectives, SLA clauses need to be achievable – or at least realistically aspired to. There is little sense in setting SLA terms that will only be achieved in an ideal world – but having some clauses that will need effort to meet will encourage improvement.


To be of any value to the organisation, an SLA needs to be documented and distributed so that all parties know what to expect, and what is expected of them. Those drawing up the terms of an SLA need to walk the fine line between woolly, sweeping generalisations and nitpicking minutiae.


In order for the terms of the SLA to be adhered to, there needs to be high level buy-in to the principles upon which it is based. All major subsections of the corporate structure need to play a part – or be offered a part – in the process of creating the SLA.

Ensuring that everyone has been given the opportunity to participate removes the later opportunity of cries of ‘we didn’t agree to this’ when – inevitably – users find that they are having to compromise.

Enforced from top to bottom

If the SLA has been agreed by management throughout the organisation, it must also be even-handedly enforced throughout the organisation. Part of the delicate task of formulating a workable SLA is that of preventing the inevitable instances of unwilling managers trying to pull rank for special treatment.

Having a very senior sponsor for the SLA – such as the CEO or other board member – allows for a point of referral and arbitration for disputes if or more usually when they occur.

A good, comprehensive SLA needs to allow for controlled exceptions to the rules – monitored and recorded for future attention – where exceptional or crisis situations occur. As part of the ongoing monitoring and refinement process that should accompany an SLA, the occurrences of these exception conditions should be closely examined.

If exceptions occur too frequently, attention will need to be given to the causes of those exceptions, and the service levels adjusted if necessary. It is not a case of ‘moving the goalposts’, but of adjusting the processes and policies to meet what is practical.


If handled properly, the formulation of an SLA is not a finite task – it is an ongoing process, periodically revisited now and evermore. Organisations change and develop. Business activities evolve. Budgets, priorities and focuses change.

The monitoring and reformulation process should be able to take account of any lessons learned since the last revision of the SLA was issued. The issuing of a revised SLA should be a smaller form of the consultation and agreement exercises when the SLA was initially drawn up.

jon m wilson Written by:

As half of the team behind, I am a serial starter of things, beginner of projects. I work in bits and in bytes, in words and paragraphs; I work in wood, metal, and paper, in fabric and in leather; I work in fits and in starts. Most of all I work intermittently and inconsistently.

Comments are closed.