In the 1990s most of the crucial business domains received help of operational IT-systems e.g. ERP, CRM, HRM helping organisations to improve their internal processes. This resulted in an exponential growth of available data causing a new problem i.e. focusing on what really matters. Here is where the Balanced Scorecard came into play. This common-sense method tries to bundle Key Performance Indicators (KPIs) from different perspectives or view-points in order to obtain a holistic view on the organisations’ activities and performance. Their aim is to make an abstraction of reality with maximum content for minimal volume of data. The iterative Pit Head KPI building method uses a scientific approach going from hypothesis generation at the strategic level over hypothesis formulation and OLAP-analysis at the tactical level till hypothesis testing through statistical analysis at the operational level. These hypotheses tests are transformed into business insights at the tactical level and are fed back to the strategic level for assessment and generation of new ideas.
By Dries Van Nieuwenhuyse, Senior Business Intelligence Consultant, Atos Origin
Lecturer Centrum Postuniversitaire Studies , EHSAL Management School
Key Performance Indicators
KPIs are metrics that are meant to trace the degree to which an organisation is realising its strategy. These metrics are featured by actuals, targets, tolerances and assessments or traffic lights. The principal aim of the Balanced Scorecard is to combine a diversity of financial and non-financial metrics to reflect all strategic issues that drive the business. The combination should give a good overview of the overall performance in function of the strategy. The construction of these indicators is an art rather than a discipline.
Top-down versus bottom-up approach
Currently different approaches are followed to define KPIs i.e. a top-down and a bottom-up approach. Both approaches have their pros and cons. The top-down approach has the advantage of guaranteeing compliance to the corporate strategy throughout all organisational levels. The disadvantage however, is that the KPIs might be isolated and not really concrete at the strategic level, sometimes without operational relevance or extremely abstract so that quantification is impossible. When no link exists with underlying data sources from the beginning, it mostly becomes tedious or even impossible to quantify the metrics afterwards. The bottom-up approach is mostly stemming from the operational work and is rather easy to quantify. Operational indicators are mostly available in operational databases or even datawarehouses. Mostly no aggregation is done and straightforward measures only characterise the operational processes. The main lack of this approach is that the KPIs are often too detailed and of no strategic relevance. Both methods relate to each other as theory versus practice.
Pit Head approach
Pit head : constantly bringing up new material
The iterative Pit Head indicator building method uses a scientific approach going from hypothesis generation at the strategic level, hypothesis formulation and OLAP-analysis at the tactical level, hypothesis testing through statistical analysis at the operational level.
1. The method works with iterative cycles that cover the entire Business Intelligence Pyramid (see fig.). At the strategic level some working hypotheses are generated during creative brainstorming sessions. Top-management is challenged in order to obtain as much suspected performance drivers as possible. Through interviews with the top and line management we obtain a view on the current · ideas on and insights into the corporate strategy · available data · known and suspect cause and effect relationships · available reports
2. Challenging ideas are taken to the tactical level where they are transformed into formal hypotheses that are ready for testing. The tactical layer entails mostly business analysts or controllers that understand the business needs and the underlying databases and their content. Business ideas to be tested are principally some suspected cause and effect relationships between company variables. Analysis of wouldbe performance drivers allows for a quick assessment of the current business knowledge within the organisation and allows for quick-wins.
3. After the formulation of the hypothesis we test them. The hypothesis testing will be done at the operational level where raw data are taken and used in statistical testing. This testing can be descriptive using commonly used BI-visualisation tools (e.g. Cognos, SAS, Business Objects, Comshare) with OLAP or reporting tools. Advanced statistical testing is performed through state-of-the-art statistical packages (e.g. SAS, SPSS).
4. The results of the tests are assessed and translated into business insights at the tactical level. These insights are further fine-tuned in order to be useful in the strategic framework into which they are fed back and assessed.
5. The obtained business insights will generate new ideas and challenges causing the cycle to restart.
Examples
The impact of the weather on the number of visitors to Kinepolis cinema theatres was suspected to be present throughout the week. The hypothesis was formulated that there would be a linear relation between the average daily temperature and the number of visitors. The visitors database was consulted and used, weather data were obtained from the internet. Linear regression was done and revealed a steady decrease in number of visitors for every increase in temperature during the weekend days but not during the weekdays. To study the split starts of performances (e.g. early: 20.00, late: 20.30) it was hypothesised that a gap would start forming between the peak arrival times for both performances. This split would reduce the peak and allow higher throughput with fewer staff. The incoming flow of visitors was counted in five minute time-buckets for both performances. For each performance the mean, standard deviation and skewness of the normal distributions were calculated and compared. The increase in gap (difference between both means) could be compared for all performances, all days in the week and for all theatres to assess the effectiveness of the split start initiative.
Assessment of the approach
The iterative method takes business ideas up and down the BIpyramid continuously. It allows to uncover and test so-called performance drivers in order to obtain an early buy-in at Clevel management. It lets seemingly unquantifiable business indicators being tested using statistics and real data. Because indicators are quantified in order to be tested it is possible to study their historical behaviour and relevance even before rollout of the scorecard. Targets can be historically calibrated and generate signals when really expected. The process also serves as preparation for the automatic feed of a Scorecard visualisation tool as it allows a quick assessment of the information that is available in the organisation. Furthermore the use of statistics guarantees a swift quantification and implementation of relevant KPIs and allows organisations to focus on what really matters. Finally quick-wins per cycle are very useful in the change management process. The tactical use of cause and effect relationships also facilitates acceptance of the introduction of the Balanced Scorecard.
