8 tips to help you improve service level management

ITSM thought leader Stuart Rance sets the record straight on service level management.

He presents actionable SLM tips, including:

Customer satisfaction is your most important metric.
Why you must keep your KPIs customer-facing.
Why you should plan improvements, even when all processes are running smoothly.

Receive a free copy of our white paper

Tips to Help  
You Improve  
Service Level Management  
by Stuart Rance  
What Is the Purpose of Service  
Level Management?  
The purpose of service level management (SLM) is to  
understand and influence your customers’ expectations  
about IT services, and to work with the rest of the IT  
organization to make sure you deliver services that meet  
these expectations. We document the customers’ agreed  
expectations in a service level agreement (SLA). Meeting  
the service levels documented in the SLA is not an end in  
itself; the SLA is just a tool you can use to help you achieve  
the real goal — meeting your customers’ expectations.  
This seems straightforward enough, but all too often poor  
service level management results in soured relationships  
with customers. In the worst cases, poor outcomes may  
include expensive legal wrangling as well as valuable  
contracts being lost when they come up for renewal.  
A Common Failing  
Service level management is probably the most  
commonly misunderstood IT service management  
process. In my experience, many organizations act as  
though SLM is all about managing SLAs. This can have  
a very negative effect on their relationship with their  
customers, which they fail to understand because all  
the metrics show that they are achieving their targets.  
If your customer is not happy but you are meeting your  
targets, it is easy for you to feel exasperated because  
your customer is being unreasonable, and for your  
customer to feel angry because their concerns have  
not been addressed. Relationships can easily turn  
very sour indeed. This situation is often described as a  
Watermelon SLA,” because it looks green on the outside  
but inside it’s all red.  
Service Level Management Has Two  
Service level management has two quite distinct cycles of activity.  
The first of these cycles can start when you are designing the service, but  
should continue throughout the life of the service. This cycle consists of  
three activities:  
. Understand. Start by making sure you understand your customers’  
expectations. Make sure you can answer questions like: Why do they  
want to use this service? How will the service create value for them?  
What will be the impact on their business if the service is not working?  
What service levels would they like you to deliver? What can they afford  
to pay for the service?  
. Negotiate. Once you have understood the customers’ expectations,  
you should compare these to your capabilities, and recognize whether  
you are able to meet the expectations and what this would cost. Then  
negotiate with the customer to find an optimal set of service level  
. Agree. At the end of the negotiation you should agree on a set of service  
level targets, and document them. These are now your agreed service  
levels, your SLA.  
These three activities are shown as a cycle because you should not  
stop after coming to an agreement. The customers’ requirements  
may change as their business changes, or as the competition or  
other environmental influences change. Your capabilities may  
also change, making it possible for you to negotiate and agree on  
improved service levels. There should be an ongoing process of  
understanding, negotiating and agreeing on service levels.  
The second cycle cannot start until the service is in use by the  
customer. This cycle also consists of three activities:  
. Monitor. You need to constantly monitor the service to ensure it  
is delivering what your customers expect. This monitoring should  
enable you to track the status of all the service level targets that  
you have documented and agreed upon with your customer.  
. Report. Create regular reports based on the data you collect  
from the monitoring. These reports should be focussed on things  
the customers care about. Of course the reports don’t need to be  
long boring paper documents, you can use a constantly updated  
dashboard, or any other communication tool that works well for  
your customers.  
. Improve. Use the information you have collected and reported to  
plan improvements to your services. Sometimes this step will  
lead to a new understanding of customer requirements and  
trigger the Understand-Negotiate-Agree loop described above.  
As with the first cycle, this second one doesn’t end. You need to keep monitoring,  
reporting and improving throughout the life of the service.  
Here are some tips to help you improve how you do service level management. You  
can make use of these tips whether you are an internal service provider, providing IT  
services to other departments within your own organization, or an external service  
provider, providing IT services to paying customers.  
1. Customer satisfaction is more  
important than meeting numerical  
The very first SLM activity in the two cycles I described above is to  
understand the customers’ expectations. When you speak to your  
customers it is unlikely that they will explain what they want in terms  
that are easy to measure. A typical customer expectation might be  
that “Downtime of the service should not have a significant impact on  
our business processes” or “Data should be protected so that we don’t  
have any embarrassing security breaches.” You should capture these  
customers’ expectations and include them in the  
You can then identify a small number of key  
performance indicators (KPIs) that you can measure  
for each of these expectations. For example, “Priority  
incidents will be resolved within 2 hours” or “We will  
achieve and maintain ISO/IEC 27001 certification.” Don’t  
have too many of these KPIs, remember that the K  
stands for Key. The idea is to measure, report, and trend  
the most important things, not everything you could  
possibly measure. If you have too many KPIs then they  
won’t help to influence behaviour or decision making.  
2. Make sure that your numerical  
targets are as SMART as possible  
It is important to have measurable service level targets in your SLA. Ideally  
every target should be SMART:  
Specific. A specific target is focussed on a clearly identified situation  
that you want to understand. For example “How quickly do we resolve  
incidents? or “How many transactions do we process each minute?”  
Measurable. A measurable target has a clearly quantified value that you  
are able to measure. This doesn’t always mean that the target has to be  
numerical. For example, a target could be “We will achieve and maintain  
certification against the ISO/IEC 27001 security standard." This may have a  
simple yes/no measure.  
chievable. An achievable target is something that you should be able to  
meet, within the constraints and resources that you have.  
elevant. A relevant target is one where achieving the target is closely  
linked to the underlying goal or purpose. For example, “Resolving priority  
incidents within 2 hours” may help to demonstrate that you are meeting  
the real goal of protecting a business process from the impact of IT failures.  
Time-bound. A time-bound target has defined times and durations over  
which it is measured. For example, “Measured over a three month period  
and reported at the end of every month."  
SMART targets enable us to clearly and unambiguously report our  
achievements, and to understand and report trends. Unfortunately, as I have  
already said, the things our customers want are often difficult to express  
as SMART targets, and so the wrong type of focus on these SMART targets  
can result in the situation I described above as a “watermelon SLA," where a  
service provider meets all of their targets while completely failing to satisfy  
their customers.  
When you report to your customer, you should include the measured values  
from the SMART targets, but you should talk about the actual customer  
expectation and experience. For example, you might say something like “Your  
expectation was that downtime of the service would not have a significant  
impact on your business process. There were two priority 1 incidents this  
quarter and both were resolved within the agreed 2-hour time. Do you agree  
that we met your expectations? This gives the customer an opportunity to  
either agree with you, or to help you understand why the SMART target was  
not right, so that you can re-negotiate the service levels.  
The important thing to remember is that the SMART targets will never  
completely capture the customer’s underlying expectations. That is what the  
service level manager is for.  
#3. Keep SLAs short and simple  
An SLA should be easy for both you and your customer to read and  
understand. It should also be easy to modify if the targets are not working  
either for you or for the customer.  
Long and complex SLAs are common when there is a contractual  
relationship between the service provider and their customer.  
This situation can be made even worse when the KPIs in the  
SLA are used to manage financial rewards and penalties.  
Service providers are encouraged to focus on meeting the  
numeric targets, regardless of whether this is in the customer’s  
interest; and customers tend to become confrontational and  
suspicious instead of facilitating the service provider’s efforts to  
help them. On one occasion I saw a service provider tell service desk agents  
to ignore priority 2 incidents and focus on priority 3 incidents, because  
the target for priority 2 incidents was safe but there was a risk of missing  
the target for priority 3. This behaviour worked as intended, at least in the  
short term, as there was no financial penalty that month, but the customer  
found out about it and was very angry. SLAs that include financial rewards  
and penalties encourage a confrontational and dysfunctional relationship  
between the service provider and their customer. Avoid using these wherever  
I have seen SLAs that include 30 or more separate documents, some of  
which were hundreds of pages long. I really don’t understand how an SLA  
like this can be used to help manage customer expectations. My advice  
would be to separate the SLA from the contract for delivery of the service.  
The contract should have contractual terms and conditions; the SLA should  
be referenced from this contract but should not be part of it. This way you  
can modify the SLA (by agreement) when you need to, and you can use it to  
help you focus on delivering services that meet your customers’ expectations.  
I do understand the need for legal language in a contract to supply services,  
but I am not convinced that we need to embed the SLA within this legal  
4. Make sure your KPIs are customer facing  
SLAs should focus on customer expectations and customer-facing KPIs, not  
internal IT-centric ones. For example, an SLA should not include KPIs like  
Number of incidents per service desk agent” or “Average time to find root cause  
of a problem." These are internal KPIs that may be important to a service provider  
but they are not relevant to the customer. Customer-facing KPIs in an SLA should  
include things like “Number of incidents not resolved within agreed times” or  
Business impact of top 5 problems” — measures like these are directly relevant  
to how well the IT organization is helping the customer meet their business goals.  
5. Use generic service levels whenever these  
are appropriate for your customers  
All the advice I have seen on service level management says that you should  
understand what your customers want and then negotiate and agree service  
levels. In fact, the Understand-Negotiate-Agree cycle that I described above  
repeats this advice.  
In practice it can be difficult and expensive to deliver individually customized  
service levels for every customer, and very few service providers do this. The  
most cost-effective approach to take is to understand what the majority of your  
customers are likely to want, in general terms, and then design your services and  
your infrastructure to be able to deliver this. Where possible you should design  
each service so that it can deliver a range of different service levels: for example,  
you might offer three different levels of availability or performance based on  
using different infrastructure to host the service, or you might offer two different  
response times for incident handling.  
If you have standard packages that you can offer your customers, then the  
negotiation and agreement are greatly simplified, because you just need to  
explain your various options to the customer and help them to decide which  
option is the best fit for their needs.  
Sometimes a customer will have a real need for a service level that is  
different to your standard offerings. You should avoid telling them  
that they have to take a standard service offering that doesn’t  
meet their need. You should put some real effort into  
understanding what the customer needs, and then working  
out what it would take for you to meet that need. Designing  
a custom solution for the customer is likely to be much more  
expensive than if they had chosen a standard service offer,  
but if you are straightforward with the customer about the  
cost of negotiating a custom service level, then this will allow  
them to make a choice based on what they can afford and  
what their real needs are.  
6. Reports should be forward looking;  
don’t just report what has happened  
An essential part of service level management is creating regular  
reports for your customers. Typically, these are created every month  
and delivered to the customer by a service level manager in a face-  
to-face meeting, but details like this obviously depend on the exact  
relationships you have with your customers.  
The service level reports you create for your customers should include  
information about your KPIs. However, it is not sufficient to just report  
the measured results for the KPI. Your reporting must help your  
customer to understand how the measurement and reporting of this  
KPI will influence how you deliver service to them. For each KPI you  
should include information about:  
l The value of the KPI for the reporting period. For example: “Average  
time to resolve P2 incidents was 4.1 hours.”  
l Whether any thresholds were exceeded. For example: “This was  
outside the agreed KPI threshold of 4 hours.”  
l The trend for the KPI. For example: “This is slightly better than last  
month and is a significant improvement on the figures from the  
previous three months.”  
l Any changes or improvements that you plan to make to improve  
the service based on this KPI. For example: “The improvements we  
have already put in place are continuing to improve the time it takes  
to resolve incidents, and we expect further improvements as we  
continue to populate the new knowledge base. This KPI will almost  
certainly return to acceptable values from next month.”  
Obviously you use the data from your monitoring to inform your  
customer about what has happened, but the most important use of  
the report is to show your customer that you understand why this  
happened and inform them about what you are planning for the  
future. You should also think about providing your customer with  
a dashboard to allow them to have real-time monitoring of KPIs  
that are especially important to them.  
Your service level reporting should be as much about the  
future as it is about the past.  
7. Plan improvements, even if you are  
meeting all your service level targets  
Some IT organizations think that if they are meeting all their KPIs then  
there is no need to make improvements. They effectively collapse  
the second SLM cycle from Monitor-Report-Improve to just Monitor-  
Report. I’ve heard two different arguments in support of this approach:  
l We are already meeting the customers’ expectations, and if we  
improve the service this could set a new level of expectation that we  
will now have to continue meeting.  
l We are an external service provider and if we improve the service  
without additional charges then this will discourage the customer  
from purchasing higher levels of service.  
I think that both of these arguments are wrong, because they  
ignore what is happening around you. Other organizations are  
making improvements all the time. Your choice is not whether to  
make improvements or stay where you are. It is whether to make  
improvements or gradually fall further and further behind your  
competition, until you become irrelevant.  
Some organizations might think that they are exempt from this  
because they have no competition, but this is never true. For example:  
l If you are an internal service provider and your customer has no  
choice other than to use your services, I can almost guarantee  
that the most senior management in your organization will have  
reviewed the possible cost and benefits of outsourcing. If you’re still  
delivering services at the cost and quality that you were 10 years ago  
then the argument for outsourcing will be irrefutable.  
l If you are an external service provider and your customer is tied to  
a long term contract, then you still need to keep improving your  
service if you wish to maintain your competitive edge and your  
reputation. Customers that are not satisfied with your services are  
less likely to purchase additional services from you. Furthermore,  
these existing customers invariably talk to other potential customers  
about the services they receive, and if they are dissatisfied this will  
have a huge impact on your ability to win new customers.  
So even if you regularly meet every KPI in your service level agreement,  
you should still deliver regular reports to your customers showing how  
you plan to improve. Maybe the improvement will simply be to change  
the KPI targets to make them a little more challenging, or maybe you  
will plan real improvements in the underlying service. Ideally you will  
offer a combination of both.  
8. You can agree on SLAs, even if the  
customer doesn’t trust them  
I have worked with a few organizations where the IT department was  
unable to put SLAs in place because the customer had had a previous  
bad experience, and thought that an SLA was simply a tool for the IT  
department to hide behind. The typical customer comment was “we want  
the service to be available all the time, we’re not prepared to negotiate a  
lower level of availability so you can have an excuse when it fails."  
In one case that I came across, the IT department did the best they could  
for a customer like this, but failure to understand the customer’s real IT  
needs through the process of negotiating and agreeing on a detailed  
SLA resulted in the catastrophic failure of a major business project after  
the organization had spent M over a 12-month period. The customer  
wanted the service to be “available all the time," and so their IT organization  
implemented high availability technology. The design meant that the  
service might stall for between 10 seconds and 2 minutes when there was  
a hardware failure. This would be a reasonable and perfectly adequate  
interpretation of “available all the time” for most customers under most  
circumstances. Unfortunately, what this customer actually needed was  
for no failure to last longer than 300 milliseconds. If the project had been  
allowed to continue, then the first time there was a hardware failure —  
even though the incident would last no more than a couple of minutes —  
the business impact would have been disastrous.  
Customers like this can be a huge problem for the service provider, because  
they are effectively denying their responsibility for understanding their own  
needs and for setting the trade-off between cost and service levels.  
An approach that has succeeded with a number of my clients is to  
acknowledge the customer’s concern and to avoid using the term  
SLA. You could, for example, go to your customer and explain that  
you are planning to make improvements to the service, and you will  
measure and report two or three things to show how effective the  
improvements are. Then ask the customer what they think you should  
measure. One organization where I recommended this has gradually  
increased the number of things they measure and report, without  
ever introducing SLAs. The service level manager is now in a position  
to really understand and influence customer expectations, and work  
with the rest of the IT organization to make sure IT services meet these  
expectations — which is the reason for wanting an SLA in the first place.  
The tips in this paper are intended to help you improve how you carry out service level  
management. If you focus on the two cycles, Understand-Negotiate-Agree and Monitor-Report-  
Improve, and follow these tips, then you will do a great job of aligning your services to the needs  
of your customers:  
1. Customer satisfaction is more important than  
#2. Make sure that your numerical targets are as  
SMART as possible  
meeting numerical targets  
Never lose focus on the really important thing:  
understanding, influencing and meeting your  
customers’ expectations.  
SMART targets are specific, measurable, achievable,  
relevant, and time-bound.  
#3. Keep SLAs short and simple  
#4. Make sure your KPIs are customer facing  
Make sure your SLAs are short and easy for you  
and your customer to understand and to change  
if necessary. Avoid SLAs that include penalties and  
rewards; if these are required, try to keep them in  
a contract away from the SLA you use to help you  
manage customer expectations.  
KPIs should be focussed on the things that matter to  
your customer – the customer facing outcomes of the  
services you deliver.  
5. Use generic service levels whenever these are  
#6. Reports should be forward looking, don’t just  
report what has happened  
appropriate for your customers  
It’s much more efficient, and cost-effective, to offer  
a small number of generic service levels to all your  
customers. If a customer really needs something  
different, then you should make sure they understand  
the full cost of providing an exceptional set of service  
Remember that the purpose of reporting is to help  
you understand, influence, and meet customer  
expectations. Don’t just report what happened but  
make sure the customer understands trends. Tell the  
customer how you plan to improve and what service  
levels they can expect in the future.  
7. Plan improvements, even if you are meeting all  
#8. You can agree on SLAs, even if the customer  
doesn’t trust them  
your service level targets  
Service improvement isn’t an optional extra. The  
alternative to improvement is not staying still — it is  
gradually becoming irrelevant. Even if you are meeting  
all your agreed service level targets you should still be  
planning how to improve the service, and reporting  
these planned improvements to your customers.  
The SLA is not an end in itself, if the customer doesn’t  
want to use SLAs then you can still understand,  
negotiate, and agree on the service you will deliver, and  
then monitor, report, and improve.  
If you follow these tips then you will deliver services that meet your customers’  
expectations. This will result in increased customer satisfaction and reduced  
customer complaints — surely that’s a goal worth a bit of effort on your part.  
Are you interested in more ITSM tips and guidance?  
We invite you subscribe to our blog.