Digital Forecasts. Publication. Published in: Business Intelligence Magazine Issue 1/2016 Pages

Transcription

1 Publication Digital Forecasts. Published in: Business Intelligence Magazine Issue 1/2016 Pages Walid Mehanna Competence Center Controlling & Finance

2 Photo: Thinkstock/iStock/alessandroguerriero Everything Under Control: For many managers today, corporate performance management with the help of tablets and other devices is a matter of course. Predictive analytics Digital Forecasts. The digitization of economic processes continues. Whole business models and value chains are changing. As a result, corporate performance management must also be realigned. By Walid Mehanna Decision-makers in all industries should prepare themselves for a paradigm shift in corporate performance management caused by digitization. From a content perspective, forecasting and driver models lie at the heart of this change, while the technology side focuses on Big Data and predictive analytics. The combination of both aspects enables companies in many industries to use new performance management approaches to overcome existing limitations. This paradigm shift has long been anticipated in corporate performance management indeed, many have been demanding it but what does it entail? In general, the shift entails replacing the reactive-analytical evaluation of historical data with modern, pro-active forecasting approaches. Through Big Data and predictive analytics, granular data such as sales figures forms the basis for automatically generated predictions which are considerably more accurate than traditionally created forecasts. Using these highly accurate predictions, companies can develop future-oriented measures in order to exert a positive influence on the expected development. People will, however, continue to play an important role as the corrector of disruptive or irregular developments, such as an 22

3 economic crisis resulting from terror attacks, and in the validation of the results calculated from the predictive models. The key to digital performance management. At the heart of the paradigm shift is the forecast as a classical performance management instrument which is used to make predictions about the expected degree of target achievement at a given point in time. While the forecasting process was often criticized in the past for being time-consuming and sometimes also «politically motivated», current technological developments give rise to new possibilities. Using stochastic models, machine learning and data mining approaches, predictions can be created more efficiently and with better results at the same time. Having said that, the use of predictive analytics in forecasting is not a completely new development. In specific use cases, e.g. operations research, this combination has already proven its effectiveness for some time. Quantified driver models. What is new is the growing change in the underlying constraints in the form of the availability of internal and external data, sufficient computing capacity for data storage and evaluation, and increasing skills for stochastic analyses among the workforce. As part of digital forecasting in companies, analysts are replacing the qualitative-theoretical cause-and-effect relationships step-by-step with data-based, quantitative-statistical correlations while continuously evaluating their validity. Hence, the new driver models are becoming the linchpins of performance management. Robust, dynamic business models provide the foundation for scenario planning, for quantifying strategic options and for evaluating business cases. By using pattern recognition in the models, companies can identify new optimization approaches (for example for logistics costs) based on alternative supply processes and continuously gather new insights into any changes taking place. In the long term, this will result in a performance management model which covers the entire value chain of the whole company. However, this is a bottom-up development, i.e. it starts with operational data models which are gradually integrated into an overall model for corporate performance management and into a comprehensive financial forecast, as shown in the diagram below. Important uses for quantified business and driver models include greater transparency concerning correlations within a company, an objective basis for decisionmaking, and seamless integration with classical performance management instruments such as the Balanced Scorecard, value driver trees and the P&L. Performance management cycles and optimization. Automated analyses shorten reaction times, permit «high frequency decisions», and lead to ad-hoc implementations of optimization measures. Prime examples for this are improving stock levels in retailing based on automatic merchandise planning or organizing maintenance contracts in the field of predictive maintenance in engineering. The ex-post and deviation-based performance management approach is expanded to include an explorative, real-time optimization approach: Algorithms bypass planned/ actual or planned/ forecast deviations to search for optimization potentials. The ongoing adjustment of value drivers leads to gains in productivity and efficiency outside of the planning and reporting cycle. The models for revealing cause-effect relationships are continuously being further developed. Automations increase efficiency while the quality and speed of management decisions are improved One-stop shop. Digital forecasting enables performance management along the entire value chain of a company in the long term. Figures are extracted bottomup from detailed models, such as for prices, and integrated into the overall model for the company and into the finance forecast. Predictive KPIs with Performance Management Relevance/ Value Driver Model Financial Forecast 1 Sales predictions based on unstructured data on trends in social media etc. 2 Fully-automated pricing based on events, personal customer information or competitor prices analyzed on a daily basis 3 Early recognition of changes in raw materials prices to optimize purchasing behavior 4 Optimization of warehousing based on current buying preferences of customers Focus on main value drivers of the company and thus on beneficial applications of predictive analytics Quantity Price Raw Materials Warehouse Staff Revenue Costs EBITDA Focus on financial top-kpis of the company Generally, P&L or value-based performance indicators are used GuV gem. 275 Abs. 3 HGB nach Umsatzkostenverfahren 1. Umsatzerlöse 2. - Herstellungskosten der zur Erstellung der Umsatzerlöse erbrachten Leistungen 3. = Bruttoergebnis vom Umsatz 4. - Vertriebskosten 5. - allgemeine Verwaltungskosten 6. - sonstige betriebliche Erträge 7. - sonstige betriebliche Aufwendungen 8. - Erträge aus Beteiligungen, davon ausverbundene Unternehmen 9. - Erträge aus Wertpapieren + Ausleihung des Finanzanlagevermögens 10. aus verbundenen UN 11. Zinsen und ähnliche Erträge, davon ausverbundenen UN 12. Abschreibungen auf Finanzanlagen + auf Wertpapieren des Umlaufvermögens 13. Zinsen und ähnliche Erträge, davon an verbundenen UN 14. nis gewöhnlicher Geschäftstätigkeit (ordentliches betriebsfremdes Ergebnis) 15. Außerordentliche Ertäge 16. Außerordentliche Aufwendungen 17. Außerordentliches Ergebnis Special Effects 18. Steuern vom Einkommen und vom Ertrag 19. stige Steuern 20. Jahresüberschuss/Jahresfehlbetrag

4 Oil field: hardly predictible prices and volumes. Photo: Flickr, author: fireboat895 24

5 significantly by digital forecasts. Increasingly, forecasting approaches are becoming established on the basis of innovations such as machine learning. In a volume-driven business without special effects they exceed the quality of manual forecasts. As a result, decisions within defined value and risk parameters are being automated based on the probability of forecast results. Among others, there are use cases in this area in merchandise planning in retailing, in algorithmic trading with financial instruments, or in dynamic pricing management in the telecommunications industry. In principle, decisions are made more quickly based on quantitative, differentiated findings and recommendations, and opportunities and risks are integrated into the analysis methods while taking crossfunctional interdependencies into consideration. This means the influence of pure expert opinions is reduced drastically. Modelling and analysis. When looking at the benefits, it can be seen that less work is needed for non-value creating activities while having more accurate results and improved business processes. The integration of opportunities and risks leads to far greater transparency concerning their potential impacts. Essentially, the quality of the data and the methods determine the quality of the results. One particularly important task lies in ensuring the validity of the external Big Data, while other decisive factors include using the right algorithms and continuously optimizing them: Thus, developing and maintaining the complex models is a critical success factor. Internal and external data. The decisive advantage of using new digital performance management instruments does not manifest itself until we combine expert and industry knowledge with the methodological skills and the entrepreneurial spirit of the interdisciplinary experts and managers involved. It is essential to work together across disciplines in order to fully tap this potential. While data scientists analyze all the data to find correlations relevant to performance management, the interpretation and processing of the findings is the responsibility of management accountants and top management. The basis for digital forecasts is raw data with maximum granularity which can be consolidated into a top KPI precisely when the information is needed. There is no longer any need for complex and time-consuming aggregations and transformations, at least not to the same extent as in the past. This means the information content remains completely intact for the analysis and there are no limits on validating the different data sources and processing steps. Companies who wish to fully tap this potential of Big Data need access to a wide variety of facts. One critical success factor is rapid availability before the data is integrated into the central data pool. This encompasses both internal and external information sources and structured and unstructured ones, such as market and customer data. The single-circuit system is increasingly becoming the established standard in accounting. This choice has significant advantages for performance management as central data storage in an integrated business suite such as SAP S/4 HANA Simple Finance is highly standardized and less fragmented. Time-consuming comparisons between external and internal accounting functions are no longer necessary, while the integrated data pool is also available for all analyses in real time and is extremely performant. A look ahead.. In the eyes of many of those in charge, digital forecasts and digital performance management might still look like a black box: Data entry and processing lack transparency or remain the domain of proven technology and statistics experts. Many managers ask themselves how they should take plausible decisions and make solid recommendations based on such results. Accordingly, it is important to develop trust in these new approaches and their results continuously and successively. For this, it is necessary to consistently build up the required skillsets among employees. Additionally, it is vital to communicate the benefits to all levels of the compa- ny and to ensure data security, data consistency and sustainable governance. Digitization is a mega-trend with the potential to trigger a paradigm shift in corporate performance management. The changes will impact upon both the senders and the recipients of performance management information. In all, this will create a completely new understanding of the performance management instruments in established use today. In future, all relevant performance management dimensions, such as products, customers, markets or resources, will be able to benefit from the added value created by Big Data and predictive analytics. However, there are still quite a few hurdles to overcome along the road to performance management with the help of digital forecasts. Although there are already some first, isolated and distinctly successful use cases in corporate practice which can take on a pilot role, viable total constructs for an integrated performance management approach are still in their infancy. What will be crucial for success are strategic orientation and the successful interaction of concept and data, methods, technologies and processes. Walid Mehanna is the Head of Business Intelligence & Big Data at Horváth & Partners Management Consultants. He started out as an information scientist, has been passionate about digital technologies for nearly twenty years and advises renowned companies in the concept design and implementation of these innovations. Alongside his project activities, Walid Mehanna is an author and lecturer at the University of Stuttgart, where he shares his knowledge and insights in this field. wmehanna@horvath-partners.com 25