Section Three:
what do these principles mean in practice?
The section above suggested seven principles that should be taken into account in
designing a system of performance measures. There are three practical considerations to
bear in mind in setting the measures themselves. Measures must be:
 | purposeful |
| clearly defined, and |
| easy to interpret. |
Purposeful measures
If the measures are to be used to maximum effect, it is important that they should be
framed in such a way as to provide the most useful information. There are a number of
dimensions to this:
 | Measures intended to drive change should focus on areas where there is most scope for
improvement – for example, where there are known to be big variations in performance. |
| Measures should be cost effective. Accurate data is essential to support evidence-based
decision making, but getting really accurate data is expensive. A balance must be struck
between the cost of collecting the data and its usefulness. All data needed for a
performance measure should also be needed as part of line management of the service. |
| A measure should seek to capture aspects of performance over which the organisation has
at least some degree of control. Otherwise it will have little impact on performance and
may be perceived to be unfair. Where organisations share responsibility for an activity,
joint indicators should be developed so that they become dependent on each other to
succeed. |
| Some areas are intrinsically difficult to measure, in particular outcomes and quality.
In such cases, proxy or partial measures should be used. For example, measures of consumer
satisfaction, the level of demand for a service, or the number of complaints may all be
taken as ‘proxies’ for the quality of a service. This is discussed further in
section four. |
Clearly defined measures
If a system of performance measurement is to generate high quality, reliable data, it
is essential that the measures are clearly defined. Again, there are a number of
dimensions to consider:
| Measures which are intended to facilitate comparisons must be consistently defined,
whether between organisations (or units within them) or over time. This may be difficult
to achieve – for example, cost data are seldom as immediately comparable as they may
seem, due to differences in the way overheads are allocated. Clear and consistent
definitions are essential to ensure valid comparisons. It also makes the information
generated more meaningful to outside audiences, for example if sickness absence is
measured in the same way across all departments. In practice this is extremely difficult
to achieve, requiring constant attention. (The Audit Commission runs a permanent helpline
for local authority performance indicators). |
| Measures should be unambiguous – i.e. any change in the measure should fairly
reflect a change in performance, rather than suggesting a change unrelated to performance
(eg through demand), or perhaps simply reflecting a different way of classifying
activities. In other words, there should be a strong causal link between performance in
the area in question and the measure. |
| Finally, measures should be robust and verifiable, so that users may be confident that
the data is reliable. |
Easy to interpret
It can often be difficult to interpret performance measures. To use turnover as a
simple example, a department will want enough staff turnover to bring on fresh blood and
provide opportunities for promotion; but not too much staff turnover since that is
expensive and loses departmental knowledge.
In this case, a target could be a range, neither too high nor too low.
Ideally it should be self-evident from the chosen measures what constitutes good
performance. But the proper interpretation of a measure will often require contextual
information. For example, to understand road traffic accident statistics and compare them
between, say, police force areas, it would be helpful to set them in the context of one of
the following:
| miles of road in the force area; or |
| miles driven per year; or |
| number of vehicles. |
The most appropriate context is rarely the "obvious" one, and may require
research to identify it.
It is also important to understand the limitations of what performance measures can
tell you. Whilst measures measure, they do not explain why performance is
improving/static/ deteriorating. This requires an analysis of the underlying processes in
a way which performance measurement by itself does not achieve. This is a particularly
important consideration when trying to compare performance between, say, different
departments.
index page | section 2 | section
4 |
