1. Introduction

Service levels and service credits are probably the most important tools used in outsourcing contracts to ensure that the supplier performs the services to the required standard. At their simplest, these tools objectively define how well the supplier needs to perform and the immediate financial consequences of failure.

While the service description in an outsourcing contract is often stated in absolute terms of what the supplier must do, the service levels specify how well the supplier must perform these services. Often this standard is less than perfect, and service levels allow the supplier to agree with the customer about what "good enough" looks like.

Service credits (or service level credits) are a mechanism by which amounts are deducted from the amounts to be paid under the contract to the supplier if actual supplier performance fails to meet the performance standards set in the service levels. While the contract may include various mechanisms to rectify poor service delivery, service credit schemes have become an established tool in outsourcing transactions to drive good performance under the contract.

This note analyses how organisations may set and use service levels in outsourcing transactions and the service credit schemes organisations may employ to encourage good performance.

  1. Service Levels

Service levels in modern outsourcing contracts are not intended to cover all aspects of the services, and do not measure all things that can be measured. Instead, they are used to focus on the most important parts of the services, and are further limited to those parts which can be objectively measured. Sometimes additional measures are included that are not linked to the service credit mechanism (i.e. there is no credit applicable in the case of failure), which are often referred to as key performance indicators or KPIs. KPIs can be viewed as another way of describing the services. As explained below, sometimes a mechanism is included to allow the customer, on giving a reasonable period of notice, to swap KPIs with service levels, so that service credits become applicable.

The appropriate number of service levels to include in an outsourcing contract will vary depending on the breadth of services. However, less is generally considered to be better than more, and ten to twenty is a good rule of thumb. Each of these service levels should be:

  • Contractually binding (i.e. contractual obligations, rather than goals or targets).
  • Capable of objective and reliable measurement.
  • Important to the customer's business objectives and focused on those parts of the service that directly impact the customer.

They should also, to the extent possible, cover the end-to-end service.

There are many ways to implement service levels, and practitioners in this field all have their favourite approaches. With no single right answers, although plenty of potential pitfalls, this note focuses on some of the options available and the market practices that have evolved in this area.

2.1 Types of Service Levels

Although service levels can be described in many ways, typically they measure the supplier's performance over a month or some other performance period. These service levels can be broken down into three categories, each of which have different characteristics that need to be defined (see box below):

  • Continuous service levels. These measure, say on a monthly basis, whether a steady state has been maintained; for example, the availability of an IT component or service. These service levels are measured on a continuous basis (albeit within certain service windows) and do not measure the completion of an event.
  • Event service levels. These measure the extent to which a series of separate events are successfully completed on time during the performance period; for example, calls answered on time or equipment installed within the relevant timeframe.
  • Sample service levels. These measure whether a sample of items meets the standard in the contract; for example, entries in an asset registry are correct or letters sent meet the required standards. These samples are used where every item practically cannot be measured in the month.

Click here to view table.

2.2 Defining the Service Levels

  1. Single performance standard

At its simplest, a comprehensive set of service levels will specify a single performance standard for each service level (e.g. the computer system is available for use by the customer 99.9% of the time) (see box below). While this approach appears simple, it is important to ensure that all terms used in describing such service levels (for example in relation to availability service levels, terms such as "system", "unavailable" and "service hours") are accurately and adequately defined and how the service level is to be measured is addressed in the contract.

  1. Compounding service levels

Service levels may . This formulation is a useful way of helping reduce the number of multiple service levels for the same topic and can be a legitimate way of creating a single service level that captures what success looks like for a particular service category. Suppliers should, however, be cautious of this formulation as a multi-tiered service level can increase the likelihood of service credits applying.

The compounding formulation is particularly useful for addressing "long tails" in service levels. For example, if a service level states that 80% of calls to a call centre are answered within 20 seconds, without including anything else in the agreement, there would be no requirement on the supplier ever to answer the remaining 20% of calls. To address this, best practice is to specify a further requirement to cover the remaining calls: for example, 80% of calls answered within 20 seconds and 100% of calls answered within 60 seconds (see box below).

  1. Continuous service level example: Service availability

The Service Level is stated as a percentage and is calculated as follows:

Service Level % = ((A − B) / A) *100


A = the number of minutes during the month, less scheduled downtime for maintenance purposes that is agreed with customer in advance.

B = the number of minutes during the month when [specify system or service being measured] is unavailable (other than during scheduled downtime for maintenance purposes).

For the purpose of this Service Level, “unavailable” means [ ]. T

he Supplier shall measure the Service Level using [ ].

The performance standard that shall be achieved by the Supplier in the relevant month for the Service Level to be achieved shall be greater than, or equal to, [ ]%.

  1. Event service level example: answering of calls to a helpdesk The Service Level is stated as a percentage and is calculated as follows:

Service Level % = (B / A) *100


A = the actual number of calls that are made to the helpdesk by users in the relevant month.

B = the number of calls that fall within "A" above that are answered by the helpdesk in accordance with this agreement within the relevant timeframe.

The supplier shall measure the Service Level using [ ].

The performance standards that shall be achieved by the supplier in the relevant month for the Service Level to be achieved are both of the following:

[80]% of all calls answered within [specify timeframe].

[100]% of all calls answered within [specify longer timeframe].

  1. Multiple performance standards

As an alternative to the single performance standard for a service level (e.g. 95% of the events were performed on time), some approaches use two (or more) performance standards (e.g. an expected level of 95% of the events being performed on time and a minimum level of 80% of events being performed on time). The second performance standard is sometimes referred to as the "critical service level", being the minimum standard below which non-performance becomes critical. The two performance levels are used for a number of different reasons:

  • Some service credit mechanisms state that a service credit is only applicable if the expected level is missed multiple times, while a single miss of the minimum level will lead to a service credit. Although this approach does not immediately penalize the supplier if it misses the higher service level standard, reducing the supplier's risk may make it easier for the supplier to accept a higher standard than it might otherwise agree to.
  • One outcome of the debate as to whether service credits are the sole remedy for service level failures is to state that within the range between the expected and minimum service levels, service credits are the sole remedy but below the minimum service level, actual damages may also be recovered.
  • The agreement may specify examples of material breach (for the purposes of the termination for material breach clause) by reference to particularly poor service level performance. If this approach is followed, such breaches may be defined by reference to minimum service levels. Among the downsides to this approach,it may lead a court to conclude that a lesser breach of the service levels is not a material breach for the purposes of termination and the supplier may only agree to a very high definition of failure.
  1. Scaled service credits

A further alternative approach is to combine service levels with the service credits using a sliding scale, so that the amount payable as a credit increases depending on how poor the performance is.

2.3 Specifying Initial Performance Standards

There are a number of ways used to specify the performance standards the supplier must meet for each service level:

  • Pre-contract performance. In an outsourcing where the functions being outsourced are currently being performed by, or on behalf of, the customer, the existing standards are the most valuable source of service level data. These should not be taken as is without further review, as often an objective of outsourcing is to improve upon, and not just replicate, the existing environment. Organisations may, however, use current performance as a starting point for defining the expected future service where major transformation of the services is not required.

Where the supplier is going to take over existing business units and/or systems (i.e. it is not providing a completely new solution for the customer), then the supplier will legitimately also be very interested in the performance of the current operation. However, if the supplier is providing a completely new solution, current customer performance is likely to be a red herring as it will have little bearing on the customer's future requirements from the supplier.

For these reasons, one of the first tasks to perform when outsourcing is to collect service level data for the current operation, and where this is not being measured comprehensively, start measuring. Three to six months' worth of sound measurements are often a reasonable baseline of service level performance, and given the timescales of putting an outsourcing contract in place, there is often time to remedy situations even where there is no current performance data. It is also important for customers to include the right in outsourcing contracts to be able to pass on data to successor suppliers on the service level performance under the contract to deal with this issue when the contract comes up for renewal and possible replacement.

  • Market standards. Many areas of service, especially at the more commoditised ends of the market, have well understood industry-standard service levels. External advisors will be able to advise on market standards, and in particular benchmarking organisations have data on what levels of service should be expected at different price points.
  • Supplier's standard offering. Although not particularly relevant for bespoke and customised service offerings, or if the supplier is taking over existing customer operations, where a supplier offers standardised services to its customers (who can benefit from the economies of scale and leverage that are possible where multiple customers buy the same service), the use of the supplier's standard offering as the basis for the contract may make sense.
  • Post-handover measurements. If the supplier is taking over an existing customer function, but there is insufficient existing data to prove the pre-handover performance of that function, the supplier may wish to ask for post-handover verification or measurement of performance. Often this would take the form of setting the performance standard based on actual performance over a measurement period. This is particularly attractive to the supplier as the measurements significantly de-risk the transaction by ensuring that the supplier is not held to a higher standard than it knows it can meet in practice.

Conversely, from the customer's perspective, post-handover measurements are very unattractive and to be avoided for a number of reasons. First, the customer would be entering the agreement not knowing what level of service to expect going forward and whether that would be acceptable. Second, the supplier is hardly incentivised to perform at its best during the measurement period, which is the time when the deal will most be under scrutiny from the broader business and its reputation can be won or lost. In addition, because of the inevitable impact of the transition, service during this period is under more threat than any other part of the term (except perhaps the final stages of termination).

If post-handover measurements are not the answer, finding reasonable alternatives may still be necessary to protect both parties' interests if the supplier really is unable to quantify meaningfully the risks associated with taking over an unknown existing operation. Such alternatives might include: variations on: introducing service level measurements over the procurement period as described above, service credit holidays (i.e. service credits not applying for an initial period), reduced service levels performance standards for an initial period and/or, back-filling using additional supplier resources to manage the practical risks of performance maintenance over the initial period. From the customer’s perspective, though, service credit holidays may not be attractive for the reason above – that this is the period when the project’s reputation is made or lost and high performance during this period by the supplier is of particular importance.

Where the service levels have not been agreed prior to contract signature, mechanisms will need to be included in the contract to address what should happen if the service levels cannot be agreed. This may include referring the matter to an appropriate third party to establish the service levels, where these can be ascertained objectively. Sometimes the only appropriate mechanism is to allow the customer to terminate the applicable part of the services, or the entire agreement, and the payment of any compensation as a result of such termination will need to be addressed.

2.4 Changing Performance Standards Over Time

  1. Like many other aspects of outsourcing transactions, specifying how things will work at one point in time is not sufficient. Change control is not the answer to all aspects of change, and the agreement needs to acknowledge and address the ongoing evolution of the services over a term that might be anything between three and 10 years long. In this light, it is desirable for the parties to consider including rules about how the performance standards will improve over time, whether on a pre-set basis or depending on the supplier’s actual performance.
  2. In identifying how the service levels might change, it can be helpful to distinguish between: a transition period, when the supplier is taking over performance of the services as-is; any transformation of the services to a new to-be state; and any ongoing evolution of the transformed services. Each of the initial and transformed stages may have different associated performance standards. Even after the final transformed state has been reached, it may be appropriate to include a mechanism for ongoing improvement.
  3. Ongoing improvement provisions can range from a statement that there will be an annual review with the parties "expecting" improvements to the service levels, but without specifying what these are, through to an automatic ratcheting based on any actual performance in the previous year exceeding the contracted standard. The latter approach should be treated by the customer with care. Unless there are other incentives for improved performance, this may actually encourage the supplier only to perform just above the contracted standard.

2.5 Measuring Service Levels

An oft-made error is to include service levels in the contract which cannot, in fact, be accurately and objectively measured. Focusing, prior to contract signature, on how the service levels will be measured (including the tools to be used) is important and the measurement approach might usefully be specified in the contract.

Issues to be considered when specifying the tools and methods include:

  • Whether the proposed tool can measure, to a sufficient granularity, to identify whether the service level is being achieved. For example, a tool that measures availability of a system by sampling must measure sufficiently regularly to identify short, but service-impacting, outages.
  • For availability service levels, it is also important that the availability of the measuring tool is itself high enough to be able to consistently measure the performance of the system being measured.

2.6 Protecting the Supplier

  1. The supplier can reasonably expect not to be penalised for service level failures that are outside its control. The supplier should in particular ensure that:
  • Either it has control over all aspects of the service level performance, or alternatively, where parts of a process are retained by the customer, that such periods are carved out of the measurement.
  • Specific acts or omissions by the customer that cause a service level failure to occur are excluded from the service level calculations.
  • Events of force majeure are excluded from the service level calculations.
  1. Other areas that the supplier should watch out for include:
  • Such small sample sizes that the actual performance standard in practice is perfect. For example, with only 19 or fewer events being measured in a month, a performance standard of 95% is actually 100%, as any failure will lead to a missed service level. Anomalies of this nature can be rectified with contractual mechanisms, such as quarterly conformance reviews (where the service level is recalculated over a three month period and averaged), to ensure the supplier has not been penalised in any month where the number of events being measured did not reach the required threshold.
  • Service levels should generally be objective and not left to the customer's discretion. The one exception to this principle is customer satisfaction surveys, which, if measured fairly, are a reasonable and useful measure of overall performance under the contract.
  1. Service levels which relate to customer satisfaction surveys are often used, with a regular sampling taken of opinions on the services. The surveys should be in a form agreed by the parties and typically will be sent to a sample of end users (which may be selected from recent users of the services) and may also be sent to key customer representatives in the business and retained organisation.

2.7 Consequences of Non-Performance

  1. The supplier will normally be responsible for measuring the service levels and should report regularly (typically monthly) on actual performance against the service levels. The contract may specify how the service levels are to be reported upon and require the supplier to carry out trend analysis on problems that are occurring. In addition, the contract may set out an applicable service credit where a service level report is not provided on time, or at all, as without a service level report, the customer does not have the necessary means to assess whether or not the relevant service levels have been met.
  2. The contract also will often specify additional activities that the supplier will perform following a service level failure, including:
  • Performing a root-cause analysis.
  • Creating and implementing a remediation plan.
  • Updating the customer of the cause and the steps being taken to remedy the fault.
  1. In addition to service credits (see below), there may be other specific remedies for service level failures in the contract, such as the option for the customer to terminate for cause as a result of particularly poor performance.

2.8 Adding and removing service levels

  1. In addition to change control, the customer may sometimes reserve the right in the contract to add reasonable additional service levels (and remove service levels that are no longer required). This may take the form of promoting key performance indicators to service levels (so that service credits start to apply) or the introduction of completely new service levels.
  2. In order for the supplier to be reasonably expected to agree to giving the customer the right to unilaterally introduce new service levels, an objective process for ascertaining the level at which such service levels are to be set will need to be agreed upon. Under this process, the service levels typically would be based on actual supplier performance over a reasonably long sample period.


  1. Service Credits

3.1 Purpose of service credits

  1. It is now standard in outsourcing agreements to include a service credit mechanism, where the customer receives a credit if actual supplier performance does not meet the contracted performance standard for a service level. The purpose of this mechanism is not to reduce the charges as a matter of course month after month (or even become a profit centre!) but instead, to focus the supplier's attention on those parts of the service that are most important to the customer and drive good performance.
  2. The threshold trigger for a service credit should be at a level of performance at the bottom edge of the range of acceptable supplier performance, so that service credits will apply only in situations where all agree that performance was inadequate.
  3. It is important, therefore, that the supplier is able to achieve the service levels and does not consider, when entering into the contract, that failure will be the norm, as this can lead perversely to likely service credits being included within the quoted charges as a hidden risk factor, increasing the charges and removing the point of the mechanism.
  4. The most effective part of service credits is arguably the attention that an account will receive from senior management within a supplier organisation if service credits apply, cutting the profitability of the account. It is also not uncommon for the personal bonuses of supplier account personnel to be tied to performance metrics that include achieving the service levels.

3.2 Service Credit Mechanisms

  1. Although the inclusion of a service credit mechanism may be the norm, there are many different types of mechanisms in use:
  • The simplest approach is for the supplier to give a credit whenever a service level is missed.
  • There can be two performance standards for a service level with a credit applicable if the lower minimum standard is ever missed or if the higher expected level is missed multiple times in a year.
  • There can be scaled service credits where the amount of service credit that applies depends on how badly the performance standard is missed.
  1. These approaches are sometimes combined with some of the other mechanisms described below, such as variable weightings and a service credit cap.

3.3 Service Credit Cap

  1. Typically, the agreement will include an overall cap on the amount of service credits that may apply in any month or year. There are evolving market standards for the level of this cap (5% to 20% of fees payable in the relevant period is the typical range), but the actual level likely to be agreed upon will depend on many factors including the aggressiveness of the performance standards, whether any failure will automatically lead to a credit and, whether the supplier will have the possibility of earning back service credits for good service in other measurement periods.
  2. In extreme cases, a customer may believe that it has obtained an above-market standard and exceptionally strong contract position, while in reality, it may be paying a premium in the charges which will counterbalance any service credits that might apply. This could actually reduce the supplier's incentive to perform while increasing the cost of the deal.

3.4 Weightings and Multipliers

  1. Given that the purpose of service credits is to focus the supplier's attention on the aspects of the service that are of most importance to the customer, service credit mechanisms often contain a weighting mechanism apportioning larger-sized credits to some service levels than others. The customer's priorities are likely to change over the term (whether because of a change in business focus or in order to respond to poor performance by the supplier in particular areas) so the more sophisticated mechanisms will include a method by which the customer can elect to reallocate the weightings between service levels upon giving reasonable notice.
  2. It is also common to see the use of "multipliers" in service credit mechanisms. The effect of these is that the service credits potentially applicable to the individual service levels will, in total, exceed the cap (although the cap will still apply). The effect is that the cap is reached when only a proportion of the service levels are failed in a particular month, rather than all of the service levels. This significantly increases the risk to the supplier of the cap being reached and consequently the multiplier will be as heavily negotiated as the cap itself. Typical outcomes range from a multiplier of 1.5 to 3.5 (e.g. if the multiplier is 2, then on average, only half the service levels will need to be missed for the cap to be reached).

Example Wording

As of the handover date, the percentage of the monthly charges that will apply as a service credit for the failure to achieve a Service Level (the "Weighting Factor") is as follows:

Click here to view table.

The Customer may, from time to time, by giving not less than [ ] months' notice, change the Weighting Factors that apply to one or more Service Levels (or change any Service Level to a Key Performance Indicator or change any Key Performance Indicator to a Service Level); provided that at no time will the aggregate Weighting Factors applicable for all of the Service Levels exceed [ ]%.

If a Service Level is not achieved in any month and the failure is not excused in accordance with clause [ ], then the service credit that will apply will be calculated by multiplying the Weighting Factor that applies in the relevant time to such Service Level, by the total charges payable in respect of the relevant month; provided that in no event, will the total service credits applicable in respect of any month exceed [ ]% of the total charges for that month.

3.5 Damages and Service Credits

Given the unenforceability of penalties in certain jurisdictions (e.g. UK) and restrictions on the customer’s ability to claim both service credits and damages in certain jurisdictions (e.g. South Africa), service credit mechanisms need to be carefully framed to ensure that they are enforceable. At the same time, there is a parallel but closely related commercial issue as to whether the amount of service credits that may apply in any given breach scenario are sufficient to reimburse the customer for its losses and, if not, whether the customer has the right to also, or in the alternative, seek damages.

Typically, the outcome of these issues in the agreement normally falls between the following extremes:

  • To treat the credit as a price adjustment reflecting the reduced value of the services received, with the customer also entitled to seek damages for breach.
  • To treat the service credits as liquidated damages and the sole financial remedy available to the customer (although the customer might also have the right to terminate for breach if the failure is severe enough).

These positions are often qualified by adopting one of the following approaches, although please note that different jurisdictions will have different limitations on which approach is capable of being adopted at law:

  • The credit is treated as a price adjustment, but if damages are also sought, the credit will be deducted from the damages.
  • The customer is entitled to make an election between the service credit or a claim for damages.
  • The customer has a period to choose whether it will either seek damages or accept the credit.
  • The service credit is treated as the sole financial remedy provided that the performance is no worse than a specified minimum threshold, but if performance goes below the minimum threshold, then damages may also be sought.
  • The service credit is treated as the sole financial remedy but if the customer terminates for breach, then damages may also be sought.

3.6 Earn-Back

  1. Some approaches to service credits include an "earn-back" concept which gives the supplier the opportunity to earn back any service credits that might otherwise be payable for poor performance. Typically these mechanisms look at overall performance over a reasonably long period, such as a year, and have conditions such as all or some of the following for a service credit to no longer apply and/or for payments to be made or credited back to the supplier:
  • The average performance over the year must exceed the performance standard for the service level.
  • For every missed service level, there must be a specified multiple of months in which the service level was exceeded.
  • The service level is not missed by more than a specified amount.
  • There were no more than a specified number of months in which the service level was missed in the year.
  1. The advantage of the earn-back approach is that it encourages the supplier to overachieve the service levels in order to have the opportunity to remedy any missed months (or counterbalance any service credits incurred later in the year). The disadvantage is that it downplays the importance of hitting the service levels each and every month, and may not match the business expectations of sustained and consistent high-quality service.
  2. A similar mechanism to earn-back that also de-risks service credits for the supplier and makes the remedy less of a hair-trigger is to include the concept of "free-misses" where the first missed service level in a period (e.g. a year) is excused.

3.7 Performance Bonuses or Incentive Schemes

  1. The supplier will often want to raise the possibility of service bonuses applying, if the service level is exceeded in the same way that a credit applies if the service level is not met. Such bonuses are sometimes referred to as service incentive schemes or service debits. From the supplier's perspective, it is reasonable to ask for the inclusion of carrots as well as sticks. From the customer's perspective the situation is not really symmetrical, and it is rare that bonuses are an attractive option to the customer for two reasons: the customer will not be able to predict what the charges will be, as it will not know whether a bonus will apply; and more importantly, the customer may well not receive a business benefit from the improved service. The service levels should have been set based on the customer's business requirements and the customer may not, therefore, receive a commensurate benefit from paying higher charges.
  2. There may, however, be two areas where bonuses or incentive schemes are appropriate and helpful to both parties. First, where the customer could receive significant benefit from the supplier meeting a performance deadline or other target that will be difficult to achieve and which cannot be guaranteed by the supplier, then offering the supplier an incentive to meet the target rather than effectively penalising it for failing to meet the deadline might be a more successful approach. Second, in business transformation outsourcings, where there are not detailed service descriptions and service levels, bonuses can play an important role as part of the overall risk-reward incentivisation mechanisms.
  3. Sometimes a service credit/debit bank is established where an account is set up into which service credits and debits are paid. This may take the form of an actual bank account or only a notional account and the mechanism and timing of the set-off of service credits and debits in the account will need to be addressed in the contract.

3.8 Missed Milestones

Most of this note has focused on service levels and service credit approaches for the "business as usual" services. Liquidated damages are also often used for the transition and transformational aspects of the outsourcing arrangements and other major projects to help ensure that deliverables are provided against the required milestone dates. Liquidated damages can be used not only for delivery of transformation or transition itself, but also for other accompanying deliverables that can otherwise become nagging issues if not addressed. Sometimes contracts are constructed on the basis that even if individual milestones are not met, the deductions for delay will not apply if the project or element of the project is finally delivered on time. The contract may also provide that additional damages may be sought by the customer if the delay exceeds a particular timeframe.

3.9 Balanced Score Cards

  1. Balanced score cards are a method used to roll-up a number of service level measurements to produce an overall outcome that measures the status of the relationship as a whole, sometimes classifying the overall status as "red", "amber" or "green". The scorecard is "balanced" because the different measures are weighted depending on their relative importance to the overall relationship. Service credits may be applicable if the relationship is "red" or "amber."
  2. A scorecard can provide a simple and helpful high-level summary report of the state of the relationship. But beyond being used as simply a reporting tool, in practice, a balanced scored card mechanism is quite rare. Primarily this is because, from the customer's perspective, there is a significant problem caused by the lack of granularity implicit in the approach, which means that even where there are significant problems in a part of the services, good results in other areas often outweigh the poor areas, so service credits rarely apply in practice. The implication is that service credits only apply where the relationship is poor overall, leading to a blunting of the leverage offered by the service credit tools.