How do I set targets in Performance Management Systems?
This 2GC FAQ looks at quantitative target-setting within the context of performance management. Quantitative targets have had bad-press in recent years, for example with regard to the UK National Health Service, but without targets measurement is of limited value. Targets are a powerful expression of the underlying purpose of measurement. Without purpose (expressed as targets) we measure for measurements’ sake, and in turn we fail to inform management activity in any meaningful way. If we know what we are trying to achieve (purpose), and how we intend to achieve it (objectives) then problems in target-setting can be reduced considerably. Use of an appropriate and systematic design framework can provide these underpinnings and help make target-setting easier.
It has been argued that the need to simplify the target-setting process has been a key driver of improvements to the Balanced Scorecard framework. Balanced Scorecard design methods have been developed to ensure that measures and targets are explicitly related to objectives recognised as actionable and important by the managers designing the system. However, even with the use of an appropriate performance management framework, target-setting remains a challenging area of system design. This FAQ aims to address some of the issues commonly encountered by clarifying what performance targets are, how they are used whilst also providing some guidelines about what constitutes an appropriate target value.
What is a target?
Before explaining targets, we need to briefly define measures. There are two forms of measures: Variables (sometimes referred to as Scalar measures) and Attributes (sometimes referred to as ‘Levels’). A variable can take any number value along a metric (units). Examples would be £x, y meters, z% or ‘customer satisfaction’. Sometimes the units require additional definition as in the case of customer satisfaction. By contrast, Attributes occur (by definition) in a limited number of states, for example we might have the binary Attribute: Pass or Fail, or the three categories: Poor Service, Acceptable Service, Delightful Service. Variables can be converted to Attributes by agreeing category limits (for example, pass marks) and Attributes can be converted to Variables via sampling. A target is a statement of the objective (intended or expected) value of the Variable or Attribute at a distinct time. Defined in this way, a target is a simple concept and is used to evaluate actual measure data; to assess performance achieved compared to performance intended or expected. However, as with many aspects of performance management the real challenge lies in providing the basis for target selection.
The Context and Rationale for target-setting
Besides the simple comparison between intended and actual performance, target-setting is used for a variety of other purposes:
- Simplified Strategic Communication. For example, the proclamation of target values for key performance measures often substitutes for a richer communication concerning strategic choices (for example communicating a target cost per transaction, rather than the description of a ‘cost effectiveness’ strategic choice);
- Communicating Benchmark Standards. Similarly, operational targets can be used as a shorthand way to describe ‘best practice’ levels for defined activities (for example, the proportion of transactions completed without error);
- Who are we? Targets for intangibles (often very difficult to measure) can be used to describe desired attitudinal characteristics, cultural attributes or more subtle goals (for example, the proportion of a workforce who advocate an organisation’s products within non-work communities).
The list is not exhaustive. This variety of potential uses, means that target-setting is often carried out concurrently with other management activities.
In our experience targets are often adopted without explicit consideration of the underlying strategic or tactical implications. The result is the adoption of targets dependent upon performance measures that cannot be collected, or which require levels of performance that cannot be achieved. In addition organisations sometimes refer to external benchmarking systems to drive their target selection, meanings that their targets drive internal behaviour that addresses generic or industry priorities rather than supporting the specific strategic intent of that organisation. Similarly, an internal marketing programme might result in ‘implicit’ attitudinal targets that run counter to broader strategic decisions made by an executive board.
Such activities and outcomes are the stuff of everyday organisational routine, but in the context of performance management system design they are unhelpful. It is important therefore to recognise how they can dilute the purpose of target-setting in its purest sense. Target-Setting is of most value when it is derived from explicit and informed consideration of underlying strategic choices and the strategic objectives - with associated measures - that are expected to address these.
Measure data, when compared with targets, alerts managers to the need for intervention. This implies that targets should reflect the subsequent need to appraise performance data. Targets should trigger an alert to managers if expected performance is not being achieved, but with sufficient tolerance that the alert fires only when there is likely an unambiguous case for intervention. Real performance management systems use multiple measures which all need to have targets set for them at an achievable level, if all the targets are firing alerts all the time because the targets are unachievable, then too much management time is devoted to investigating WHY this is the case and the utility of targets to warn managers about significant issues will be diminished, in a similar way to the story about the child who cried ‘wolf’ too often!
Choosing a Target Value
As should be obvious from the preceding discussion of measures, the actual target depends upon the type of measure used, but three general categories of target apply:
- Where we have multiple or single variables (scalar measures) targets are usually Thresholds. i.e. we need to reach a fixed value on linear scale – one side of the threshold is OK, the other side is not;
- Limit based targets (i.e. 100% or 0%) are usually operational or aspirational. Operational limits may describe a required operating pattern (“... this is how we want you to behave.”) Aspirational targets may be impractical to achieve (“... we are the World’s Favourite Airline”), but may define ‘worthy’ goals;
- Lastly, implicit in the discussion of multiple measures are Rule based targets. i.e. we need x out of y to hit a limit or threshold type target for overall target achievement.
The ultimate informants of these thresholds, limits and rules are what the organisation is trying to achieve; strategically, operationally or both.
Targets derived from the long-term stretch goals will need to be broken down into manageable sub-goals corresponding with the review cycle of the performance management system. If the behaviour of the measure variable is anticipated to be linear over time, then simple division achieves decomposition of long-term targets into annual or quarterly targets. However, not all measure variables will behave in this way. For example, if the measure were project milestone achievement, then discrete step changes in the measure variable would occur when each milestone is achieved. By contrast, a measure variable might be expected to change geometrically after a period had elapsed. The various outcomes of quality improvement programmes would likely behave in this way, with little tangible return on investment in improvement activity for some time after the initial investment.
Another challenge is in decomposing a long-term strategic objective that is a binary attribute, such as service quality, into an interim outcome target. Here it may be the case that the only possible decomposition is into a series of ‘enabling’ milestones much like backward scheduling in project management.
Knowing enough to set a sensible target is particularly difficult when there is no baseline, and this is often the case when an organisation adopts a ‘new’ measure. In 2GC’s experience, about half of all measures selected for corporate performance management systems are new. In such cases it may be necessary to first measure current performance for some months, before it will be possible to establish what a realistic target might be.
Frequently and infrequently reported measures within the same system
Often there is a need to reconcile measures that have long reporting cycles with other more frequently reported measures that can be used as ‘proxies’ for these values. The need for this substitution is that without frequent data from the proxy measures, insufficient information may be available for the management team to intervene on a timely basis. Under these circumstances, the long reporting cycle data can be used to ‘calibrate’ the target values set for the more frequent proxy data.