Designing effective KPIs

In a couple of posts I highlighted the damage that poorly considered KPIs and incentive payments can cause either to the organisation or its customers:

• The normalisation of deviant behaviours looked at the effect of bonuses in the on-going CBA scandal.
• What you measure is what you get describes the effect of KPIs on behaviour.

This post fills the missing link and discusses the practical challenges of creating effective KPIs.

Key Performance Indicators (KPIs) exist to influence decisions and actions; effective KPIs motivate people towards taking valuable, and useful, actions and decisions. Each KPI is a measure of how well a fundamental part of the project (or organisation) is progressing towards achieving its goals. The elements of a KPI are:

• Key = something that is important, essential, fundamental.
• Performance = the execution or accomplishment of work
• Indicator = a measure, and record of variations

The specific purpose for each KPI is to communicate a relevant summary of the current situation to a particular person, or group; giving an indication of how effectively a particular element of the project (or work) is achieving its objectives. Because the KPI is an ‘indicator’ it does not have to be all encompassing, or provide all of the information about the activity. The purpose of a KPI is to highlight if and when more investigation is needed; they do not replace everyday ‘project controls data’ and other management information.

The challenge with KPIs is to set measures that provide indicators of potential problems in sufficient time to allow investigating and action. The purpose of most projects is to create value through the realisation of benefits; unfortunately this ‘real measure’ only happens after the project is finished. So whilst tracking benefits realised is important, the information lags behind the actions that affect the outcome. Other leading indicators are needed that focus on the probability of generating value during the course of the work (which is more complex than simply measuring time and cost performance).

The way to design effective KPIs involves six simple steps:

1. Understand your audience and tailor specific KPIs for different levels and groups within the project and the project’s stakeholder community. Detail should decrease as you move up that structure, what’s useful to a team leader is information overload for a sponsor.
2. Be clear and concise. Each KPI should be designed to deliver a message that will instigate one of two decisions; either ‘do nothing’ or ‘investigate’! The KPI’s job is to tell you one of these three things (any more information and it is not an ‘indecator’):
   1. Things are looking bad – investigate and fix
   2. Things are looking good – investigate and learn
   3. Things are OK – do nothing.
3. Make the KPI understandable. The KPI is an indicator of how well specific work is being done, or accomplished; being clear about precisely what work and what goals is critical. This means the KPI has to:
   1. Be well written;
   2. Contain one clear measure;
   3. Set realistic targets;
   4. Be time framed;
   5. Define how the data will be tracked.
4. Balance the KPIs across the performance window:
   1. Input KPIs – measure the quantity and sometimes quality of inputs to the project.
   2. Process KPIs – measure the quantity and sometimes quality of the work required to produce certain expected outputs.
   3. Output KPIs – measure the quantity and sometimes quality of the goods or services created.
   4. Value KPIs – measure the quantity and sometimes quality of the results achieved through the delivery of the goods and services eg, benefits realised.
5. Use both types of KPI:
   1. Target KPIs focus on achieving a specific measure (pass / fail), usually within a time frame, eg, units delivered per week.
   2. Directional KPIs measure tends. With many KPIs the precise number is less important than the trend. For example, “Number of days lost to staff sickness” [per month]. Here the exact number of days is not that useful as we can’t control this, however if the trend is rising we can investigate and take action accordingly.
6. Test and fine tune the KPIs, make sure you are getting the results you want. As both of the referenced posts have demonstrated, it can lead to disaster if you simply design, then implement, a KPI as a way to allocate bonuses without fully understanding if and how it can be ‘gamed’ or how it will affect morale, or any other unforeseen outcomes. Therefore:
   1. Allow some lead time to check that everyone understands the KPIs, if the outcomes being measured are reasonable and the data is easy to collects and accurate.
   2. Trial the KPI to make sure it is driving the behaviours you desire.

Finally, the characteristics of good KPIs are:

• Simplicity. The metric name should be less than 5 words and the calculation is easily described in under 10 words.
• Comparability. The measure is comparable to other time periods, sites, or segments.
• Incremental. A rate or ratio is better than an absolute or cumulative value.

Some good KPIs include:

• The accident (and ‘near miss’) rate on engineering and other ‘hard hat’ projects, a low rate indicates a safe environment which means a clean, well managed and well planned workplace.
• Performance measures such as the number of activities completed within 5% of the estimated time (the workers cannot control the start but can control the flow of work once started).
• The number of open issues (and the trend), or the number of issues that remain open after a ‘reasonable period (say 2 weeks).
• Quality measures.

A final thing is to remember setting two or three effective KPIs and using them effectively across all projects is better than a scattergun approach. You know you have too many KPIs when you hear people saying things such as the “top KPIs” or “most important KPIs”. Keep them simple, consistent and rigorous for the maximum benefit.