Where is streaming analytics mainly used?

Chapter 4: Organization & IT Big Data Analytics in Controlling: Application areas, advantages and implementation using the example of SAP HANA n Big Data Analytics represents an expansion of the analysis spectrum in Controlling, which supplements classic business intelligence with advanced analysis methods. n In-memory systems represent an essential technological basis for big data analytics. They can process mass data in real time and enable both timely decisions and reliable forecasts. n This article considers the potential benefits of in-memory big data analytics for controlling, but also the associated technological challenges. n Using the example of the in-memory platform SAP HANA, streaming analytics and predictive analytics 2 approaches and supplementary use cases for big data analytics in controlling are dealt with. Contents page 1 From data growth to big data .......................................... ....... 163 2 Big Data Analytics expands the analysis spectrum ....................... 163 3 Big Data Analytics with in-memory systems. .............................. 164 3.1 Characteristics of in-memory systems ........... ........................... 164 3.2 Possible uses and potential benefits .................. ................. 165 4 Big Data with SAP HANA .......................... ...................................... 166 4.1 In-Memory-Platform SAP HANA ... ................................................. 166 4.2 SAP Streaming Analytics .............................................. .................. 167 4.3 SAP Predictive Analytics ........................... ...................................... 168 4.3.1 Properties ....... .................................................. ....................... 168 4.3.2 Modules ...................... .................................................. . ................. 169 4.3.3 Areas of application ............................ ......................................... 170 5 Recommended actions for controlling ... ............................. 171 6 References .................. .................................................. ...... 172 161 Big Data Analytics in Controlling n The authors Dr. Christian Bischof, Associate Professor of Applied Computer Science with a focus on business information systems and Head of International Supply Management at FH JOANNEUM, Institute Industrial Management in Kapfenberg. Daniela Wilfinger, Researcher and Lecturer at FH JOANNEUM, Institute Industrial Management in Kapfenberg / Austria. 162 Organization & IT 1 From data growth to big data The acceleration of technological progress based on innovative digital technologies has led to exponential growth in the volume of data in recent years. This development will continue to grow. Data from the Internet and social networks are just one source of this increased volume of data. The main drivers for the data explosion are mobile apps, cloud computing, but above all the increasing sensor-based networking of consumer goods and capital goods or machines in the context of embedded systems and the Internet of Things (IoT) .1 According to an estimate by IDC, this will become a global one generated data volume of 163 zettabytes by 2025. At this point in time, an average digitally networked person will - consciously or unconsciously - interact with the corresponding devices and systems 4,800 times a day, this corresponds to an interaction every 18 seconds.2 Big data is therefore increasingly becoming all areas of human life and thus also the performance area and record all functional areas of a company. Up until now it was mainly internal and structured data that were processed and analyzed with the help of enterprise resource planning systems (ERP) and business intelligence systems (BI), but in the future it will also be necessary to export semi- and unstructured data to include different internal and external data sources. This requires cost-effective, location-independent, high-performance and real-time storage, processing and provision of this data.3 However, conventional technologies such as relational databases and established system architectures with their separation into transactional (ERP) and analytical (BI) systems quickly reach their limits. This is where the concept of Big Data comes in by providing technologies and tools with the help of which the acquisition, storage and analysis of large unstructured, semi-structured (mixed) and structured data volumes is technically possible and also economically interesting. 4 2 Big Data Analytics expands the analysis spectrum For controlling, Big Data Analytics represents a supplement or expansion of the established analysis spectrum, which in many companies is primarily located in the descriptive and diagnostic area. In addition to periodic standard reports to answer the question “What happened?”, The 163 Big Data Analytics in Controlling 1 cf. Seufert, 2016, p. 40. 2 cf. Reinsel / Gantz / Rydning, 2017, p. 3ff. 3 See Schön 2018, pp. 413ff. 4 Cf. Dittmar, 2016, p. 56ff. Context of Business Intelligence Tools for diagnostic questions in the form of multidimensional ad-hoc evaluations to provide answers to the question “Why did it happen?”. In addition, the use of statistical methods for pattern recognition in the traditional BI context under the term data mining has been established for many years.5 This is where the intersection with big data analytics begins, which focuses on exploratory issues. For this purpose, mathematical and statistical formulas and algorithms are used with the aim of generating new information and being able to calculate forecast values. Driven by innovative digital technologies and the larger database, new approaches have been established under the terms “Predictive Analytics” and “Prescriptive Analytics”, which primarily aim to answer the questions “What will happen?” And “What needs to be done?”. 6 In the context of predictive analytics, advanced methods such as decision trees, neural networks or support vector machines are used under the term of machine learning in order to recognize patterns in historical data and to be able to derive automated mathematical models from them to predict future behavior Combined with semantic rules, simulations or optimization processes, systems can transform the results of the analyzes directly into recommended measures. Since the systems act prescriptively here, ie make statements about “what needs to be done”, one also speaks of prescriptive analytics.8 3 Big data analytics with in-memory systems In order to be able to use the advanced analytical methods of big data analytics in controlling , the existing IT landscape needs to be expanded to include in-memory systems that enable data to be recorded and evaluated in real time. In the following, the essential technological properties and the resulting possible uses of in-memory computing in the context of big data analytics are examined in more detail. 3.1 Characteristics of in-memory systems The steadily growing amounts of data as well as the demand for a timely supply of information place high demands on the performance and economy of the information technologies used. These can be achieved with the currently still largely prevailing separation of transactional and 164 Organization & IT 5 cf. Felden, 2016, p. 2. 6 cf. Iffert et al., 2016, p. 9f. and ICV 2016, p. 29f. 7 Cf. Chen et al., 2014, pp. 1165ff. 8 Cf. Dursun / Haluk, 2013, p. 361. Analytical systems are not or only insufficiently fulfilled. 9 Was this separation an indispensable compromise due to hardware and software-related restrictions in order to be able to write as well as read and evaluate data with high performance, Today, in-memory systems provide a new technology with which both operations can be carried out in real time in a database. With this approach, data is no longer stored on hard disks, as was previously the case, but is kept permanently in the main memory of the in-memory system.10 The much faster access times and a column-oriented database management system optimized for the main memory mean that very large amounts of data can be searched much faster , processed and prepared.11 The use of in-memory systems not only increases the speed of data processing and thus the availability of data. This technology also makes it possible to fundamentally change and optimize business processes and thus make them simpler, more efficient and / or more adaptable with regard to changes in the business environment.12 3.2 Possible uses and potential benefits The use of in-memory systems does not make data more persistently stored in layers, but only managed logically. The classic data warehouse architecture is thus virtualized, and the different data layers and aggregates that were previously used mainly for performance reasons are becoming obsolete. Instead, during analyzes, the data is selected, calculated and, if necessary, aggregated without specifically storing it physically.13 The availability of transactional and analytical data in a database can avoid information latency and thus delayed provision of analytical evaluations by controlling. 14 This enables prompt decision support, which is one of the main potential benefits of in-memory systems in controlling. This is the result of an empirical survey of controlling experts on this topic. Accordingly, the decisive advantages of in-memory technologies are the acceleration of analysis processes and the prompt provision of information relevant to decision-making, as well as the establishment of real-time simulations and the immediate adaptation of these 165 big data analytics in controlling 9 cf. Knabke / Olbrich, 2016, p. 189 10 Cf. Gröber et al., 2018, p. 50. 11 Cf. Prassol, 2015, p. 363ff. 12 Cf. Marden / Olofson, 2018, p. 1. 13 Cf. Knabke / Olbrich, 2016, p. 195. 14 Cf. Kaum et al., 2015, p. 10. Scenario models for the changed corporate environment. Furthermore, in-memory technologies are seen as the prerequisite for the implementation of complex forecast models and the (resulting) improvement of forecast results. The changes induced by in-memory technologies are estimated to be the highest in the areas of planning and forecasting as well as (management) reporting.15 A study by IDC, which examines the potential benefits of in-memory systems using the example of SAP, comes to similar results HANA raised. Accordingly, the use of this technology leads to better business results, partly because of strategies based on real-time analyzes and the provision of data through self-service analyzes.16 4 Big Data with SAP HANA 4.1 In-Memory Platform SAP HANA The in-memory platform SAP HANA was first introduced in 2010. The central element of this technology is a database that is completely relocated to the main memory. The data is only physically backed up for archiving and restoration purposes. In addition, data is not stored in a row, but in a column. The logical structure of a data record is resolved during the storage process and the content is stored in columns. This means that data operations that only access selected columns of a logical table can be carried out much faster than would be possible with a line-oriented approach.17 In addition, summary tables no longer have to be set up. Instead, the data can be processed in real time with full granularity at any time. In combination with column-oriented storage, this leads to a significant reduction in the database volume.18 Many of the processes required to implement Big Data Analytics are already integrated in SAP HANA. This includes, for example, text analysis, the processing of spatial and geospatial data and the processing of streaming data.19 In addition, other analyzes are made available in HANA-based solutions, such as methods for association or time series analysis as part of the Predictive Analytics Library (PAL). Finally, procedures and platforms from third-party providers are also supported.20 166 Organization & IT 15 Cf. Gröber et al., 2018, p. 51f. 16 Cf. Marden / Olofson, 2018, p. 2. 17 Cf. Koglin, 2016, p. 55f. 18 Cf. Eilers 2016, p. 186. 19 Cf. SAP n.d., n.p. 20 Cf. Mattern / Croft, 2014, p. 105. In the following, with SAP Streaming Analytics and SAP Predictive Analytics, two selected approaches to Big Data Analytics will be examined more closely using the example of SAP HANA. In the streaming analytics sector in particular, SAP is considered to have a leading position on the market.21 4.2 SAP Streaming Analytics The term streaming analytics subsumes platforms that collect streaming data and can use this as a basis to create analyzes. Such platforms are also able to immediately execute technical operations or process steps on the basis of rules or models for machine learning, if this is necessary.22 In traditional databases, data is initially stored for the purpose of analysis. SAP Streaming Analytics makes it possible to connect different data sources via input adapters to the Smart Data Streaming Engine, which is integrated in the SAP HANA platform.23 In this way, data from a large number of sources is transferred to the in real time without prior intermediate storage SAP HANA system.24 The main potential benefit of SAP HANA for controlling is the combined analysis of different internal and external data sources in real time with the help of the integrated Smart Data Streaming Engine (see Fig. 1). SAP HANA with Smart Data Streaming Sensors Social Media Market Data Transactions Input Adapter SAP Smart Data Streaming Engine SAP HANA Output Adapter Warning Messages Dashboards Databases Applications Fig. 1: Architecture and functionality of SAP Streaming Analytics25 167 Big Data Analytics in Controlling 21 Cf. Gualtieri et al. , 2017, p. 3. 22 Cf. Hovsepian, 2018, oS 23 Cf. Pledereder, 2018, no p. 24 See Gualtieri et al., 2017, p. 3. 25 Source: Pledereder 2018, n.p. Incoming data streams are processed, examined for patterns and trends, missing values ​​identified and correlations monitored. Subsequently, information is provided in real time in the form of live dashboards, warning messages are generated or transactions are initiated in the ERP system. In addition, the data can be stored in the SAP HANA database, but also in NoSQL databases for further statistical evaluations.26 In this way, the concept of complex event processing can be established and thus an essential contribution to the implementation of real-time Enterprise ”. 27 A key area of ​​application for streaming analytics in the area of ​​controlling is production controlling. More and more machines are digitized - either originally or by means of retrofitting - and thus provide extensive sensor data. As part of automated production controlling, there are extensive options for monitoring and optimizing the entire production process. These include, for example, the predictive maintenance of the machines, the timely planning of necessary repairs, the automatic monitoring of criteria that have a negative effect on the quality of the end product and / or the order costs as well as the early identification of defective products as a result of a malfunction. For this purpose, the data from SAP Smart Data Streaming is recorded, filtered, transformed and continuously monitored. If certain thresholds for critical machine parameters are exceeded, notifications are generated by the system or events (maintenance orders) are triggered in the ERP system. The storage of sensor data in SAP HANA represents the database for statistical analyzes, for example to identify error patterns. Another potential lies in the linking of sensor data with order data and quality management data.28 4.3 SAP Predictive Analytics 4.3.1 Properties With SAP Predictive Analytics offers SAP offers methods and instruments for statistical data analysis and evaluation, with the help of which the construction of predictive models can be accelerated and (partially) automated. Thanks to the high level of integration of SAP Predictive Analytics, companies should be able to use the potential of Big Data more within their existing SAP landscape. In addition to classic on-premise use, it is also possible to implement predictive models based on the cloud.29 168 Organization & IT 26 Cf. Du, 2015, p. 5. 27 Cf. Gualtieri et al., 2017, p. 3 28 See Bauer, 2017, oS 29 Cf. Bakhshaliyeva et al., 2017, pp. 77f. SAP Lumira SAP AutomatedAnalytics SAP Expert Analytics SAP HANA Integration (PAL) SAP Predictive Analytics Required analytical knowledge Field of competence of the controller Fig.2: Target groups and product portfolio of SAP Predictive Analytics30 The target groups of SAP Predictive Analytics are very broad and range from business users with little or no analytical knowledge to business analysts and data scientists with expert knowledge in the area of ​​predictive analytics. This also covers the entire current and future field of competence of the controller with regard to data analysis and reporting (see Fig. 2). 4.3.2 Modules SAP Lumira does not have any functionality in the area of ​​prediction, but it does allow access to the SAP HANA database and thus data analysis in real time. Combined with interactive dashboards, users in the specialist departments as well as controllers are able to create interactive ad hoc evaluations and analysis applications without the support of the IT department.31 SAP Automated Analytics provides various data mining methods that can be used with the help of a graphical user interface enable a quick and user-friendly definition of analysis models and algorithms. By providing an automated data preparation and modeling process with predefined algorithms, automated analytics is primarily suitable for standardized analyzes in controlling that do not require in-depth mathematical knowledge. These analyzes can be carried out online, with direct access to the SAP HANA database, as well as offline.32 SAP Expert Analytics supplements the visualization options of SAP Lumira with a forecast function. As part of process-oriented modeling, numerous data mining algorithms are used, from cluster analysis to 169 big data analytics in controlling 30 Based on Bakshhaliveva et al., 2017, p. 79. 31 Cf. Plederer, 2017, o.S. 32 Cf. Kraus / Kerner 2018, p. 16. neural networks. Although it is supported by a graphic process design tool, the required parameterization of the algorithms requires a basic understanding of mathematics and statistics.33 In addition, individual algorithms can be implemented with the help of statistics programs and executed online in real time or offline.34 In addition, predictive models can also be used directly in SAP HANA and therefore very high-performance implemented. Various libraries are available to the user for this purpose, above all the SAP Predictive Analytics Library (PAL). The functions and algorithms provided within the framework of the PAL are triggered by SQL procedures in SAP HANA in order to carry out, for example, regression, association and time series analyzes as well as analyzes of social networks.35 4.3.3 Areas of application A promising area of ​​application of predictive analytics in the field of Controlling is the forecast. With the help of innovative technologies, there are new possibilities to both increase the quality of the forecast and to make its creation more efficient.36 This is achieved by replacing qualitative-theoretical cause-effect chains with data-based stochastic forecasts and deterministic models.37 The forecasts relate to parameters from the corporate environment, such as sales forecasts or raw material prices, but also within the company, such as downtimes in production. These forecasts, created with the help of SAP Predictive Analytics, then serve as input for defined driver models.38 The driver tree created in this way integrates the various sub-plans “bottom-up”, taking causal relationships and effect relationships into account. In this way, an integrated value driver-based planning model is created that reveals the interdependencies (and also the time lag) of the individual elements and thus allows a time-related perspective. Based on this concept, the controlling department can create forecast models and carry out additional sensitivity and “what-if” analyzes.39 Due to the large amount of resources required, such calculations can often only be carried out overnight using traditional technologies, which largely excludes short-term adjustments and recalculations. The use of SAP HANA now also allows short-term recalculations in 170 Organization & IT 33 Cf. Bakhshaliyeva et al., 2017, p. 83. 34 Cf. Lauer et al., 2018, p. 37. 35 Cf. Bakhshaliyeva et al., 2017, p. 83. 34 Cf. Lauer et al., 2018, p. 37. 35 Cf. , 2017, p. 366. 36 Cf. Mehanna et al., 2015. P. 29. 37 Cf. Schell et al., 2017, p. 222 and Mehanna et al., 2015, p. 30. 38 Cf. Mattern / Croft, 2017, p. 196. 39 Cf. Mehanna et al., 2018, p. 41. If necessary, it offers improved decision support through timely, up-to-date information. In addition, there is the possibility of establishing new business processes that monitor the forecasts and plans in real time, both on the basis of rules and with the help of statistical methods, look for outliers, carry out ongoing impact analyzes and, if necessary, alert the employees concerned or initiate measures themselves. 40 5 Recommendations for Action for Controlling Both conceptually and technologically, Big Data Analytics has now reached a level of maturity that enables a high degree of diffusion and application in controlling. The potential benefits derived from this are largely undisputed, but also the fact that Big Data Analytics will permanently change the methods, systems and processes in controlling. For this reason, from the controlling point of view, it is urgently necessary to deal with this topic proactively.41 In-memory systems are not only an essential technological component, they can also be a first essential step towards the implementation of big data analytics in controlling be. Due to the upcoming release change of the SAP ERP system used by many companies to SAP S / 4 in the next few years, there will be an automatic technological leap to the in-memory platform SAP HANA. However, an early and continuous involvement of controlling is required in order to integrate the functions contained in SAP S / 4 in the context of big data analytics into the existing controlling concept and to use the possibilities of advanced analysis methods accordingly. Even if in-memory systems provide essential functionalities for big data analytics, it is not advisable to keep all data in the main memory for a long time. A flexible concept for big data analytics is therefore required, which stores data in memories with different “temperatures”. Data for real-time applications and analyzes are held in "hot storage", while bulk data for statistical analyzes, for example, are stored in "cold storage". The task of controlling in this context is to classify the data accordingly against the background of current and future reporting and analysis requirements. Ultimately, the targeted use of big data analytics also requires an expansion of the controller's technical competence profile. It is true that the controller will not take on the tasks of the data scientist in the near future, 171 Big Data Analytics in Controlling 40 Cf. Mattern / Croft 2017, p. 196. 41 Cf. Grönke / Kirchmann / Leyk, 2014, p. 63ff. however, skills in this environment will be essential. These include an understanding of the basic technologies, the essential data structures and data flows in the company, but also the relevant analytical and statistical models. In this way, the controller can use Big Data Analytics to strengthen his role as a business partner for management.42 6 References Bakhshaliyeva / Chen / Dommer / Semlenski / Schmedt / Schulze / Wilczek, SAP Predictive Analytics: Predictive Analytics with SAP, 2017. Bauer , Process controlling with SAP, 2017, https://blog.orbit.de/2017/09/19/process-controlling-mit-sap/, retrieved on October 8, 2018. Chen / Chiang / Storey, Business Intelligence and Analytics: From Big Data to Big Impact, in MIS quarterly, vol. 36, 2012, no. 4, pp. 1165–1188. Du, SAP HANA Smart Data Streaming and the Internet of Things, 2015. Dursun / Haluk, Data, information and analytics as services, in Decision Support systems, vol. 55, 2013, no. 1, pp. 359–363. Dittmar, The next evolutionary stages of AIS: Big Data, in Gluchowski / Chamoni, Analytical Information Systems, 2016, pp. 56–65. Eilers, SAP S / 4 HANA: New functions, application scenarios and effects on financial reporting, in Group Controlling 2020, 2016, pp. 183-200. Felden, Predictive data analysis as a driver of business activity: Advanced and Predictive Analysis, 2016, http://www.sigs.de/publications/bi/2016/Predictive_AdvancedAnalytics/felden_BIS_OTS_2016.pdf, accessed on October 8, 2018. Gröber / Schlecht / Esch / Gleich, In-Memory-Technologie, in Controller Magazin, May / June 2018, pp. 50–53. Grönke / Kirchmann / Leyk, Big Data: Effects on instruments and organization of corporate management, in Gleich / Grönke / Kirchmann / Leyk (Ed.), Big Data and Controlling: Requirements, Effects, Solutions, 2014, pp. 63–82. Gualtieri / Sridharan / Hoberman, The Forrester Wave: Streaming Analytics, Q3 2017, https://www.forrester.com/report/The+Forrester+Wave+Streaming+Analytics+Q3+2017/-/E-RES136545, 2017, Access date October 8, 2018. Horváth / Aschenbrücker, Data Scientist: Competition or Catalyst for the Controller ?, in: Gleich / Grönke / Kirchmann / Leyk (Eds.), Big Data and Controlling: Requirements, Effects, Solutions, 2014, pp. 47–62. 172 Organization & IT 42 See Horváth / Aschenbrücker, 2014, p. 61. Hovsepian, The Competitive Advantage of Streaming Analytics, 2018, available online at: https://www.forbes.com/sites/forbestechcouncil/2018/05/ 11 / thecompetitive-advantage-of-streaming-analytics / #, accessed on October 17, 2018. ICV Internationaler Controllerverein, Business Analytics - The way to data-driven corporate management, 2016. Iffert / Bange / Mack / Vitsenko, BARC user study: Advanced & Predictive Analytics: Key to future competitiveness, 2016, http://www.bi.soprasteria.de /docs/librariesprovider56/default-document-library/barc-advancedund-predictive-analytics-2016.pdf?sfvrsn=2, accessed on October 9, 2018. Kaum / Töller / Wende / Moers, Influence of In-Memory Technology (SAP HANA) and Simple Finance on Controlling, in Controller Magazin, September / October 2015, pp. 10–15. Kraus / Kerner, SAP Analytics Cloud, 2018. Koglin, SAP S / 4 HANA, 2016. Knabke / Olbrich, Basics and Application Potentials of In-Memory Databases, in Analytical Information Systems, 2016, pp. 187-203. Lauer / Merkt / Müller / Tschimmel, SAP Lumira, Das Praxishandbuch, 2018. Marden / Olofson, SAP HANA Customers Deliver Real-Time Data-Driven-Innovation with Significant Business Value and Return on Investment, IDC-Whitepaper, 2018, https: / /www.sap.com/documents/ 2016/11 / 60ad4c8b-937c-0010-82c7-eda71af511fa.html, accessed on October 18, 2018. Mattern / Croft, Business Cases with SAP HANA, 2014. Mehanna / Müller / Tunco, Predictive Forecasting and the Digitization of Corporate Management, in IM + io - Trade Journal for Innovation, Organization, 2015, no. 4, pp. 28–32. Mehanna / Tatzel / Vogel, Business Analytics in Controlling, Controlling: Journal for Success-Oriented Corporate Management, Special Edition 2018, pp. 39–45. Prassol, SAP HANA as an application platform for real-time business, in HMD Praxis der Informatik, H. 3, 2015, pp. 358–372. Pledereder, SAP HANA Streaming Analytics - Recognizing a Leader, 2017, https://blogs.saphana.com/2018/02/07/sap-hana-streaming-analytics-recognizingleader/, accessed on October 8, 2018. Reinsel / Gantz / Rydning, Data Age 2025: The Evolution of Data to Life-Critical - Don't Focus on Big Data; Focus on the Data That's Big, IDC White Paper, 2017, https://www.seagate.com/www-content/our-story/trends/files/Seagate-WP-DataAge2025-March-2017.pdf, accessed on 7.10. 2018. 173 Big Data Analytics in Controlling SAP, knowledge gained with the highly developed analysis methods from SAP HANA, no year, https://www.sap.com/austria/products/hana/features/advanced-analytics.html, retrieved on October 17, 2018. Seufert, Digitization as a Challenge for Companies: Status Quo, Opportunities and Challenges in the BI & Big Data environment, in Fasel / Meier, Big Data, 2016, pp. 39–57. Schell / Schmid-Lutz / Schocke / Stockrahm / Zinovieva, Industry 4.0 with SAP, 2017. Schön, Planning and Reporting in BI-Supported Controlling, 3rd Edition 2018. 174 Organization & IT Predictive Planning in SMEs: Advantages and Implementation in 5th Edition Steps n In medium-sized companies, there is an increasing need to continuously update planning during the year. Forecasting is therefore gaining in importance in corporate planning. n The development of a data-driven company offers controllers the opportunity to use modern technologies even without IT background knowledge. n The prerequisites for the successful use of predictive planning methods and forecasting are clean data management, central data storage and integrated corporate planning. n Dealing with increasing market dynamics and volatility requires flexible teams and an open “trial and error” culture. n Using the example of the Corporate Planner software, the article shows the advantages of predictive planning and its implementation in 5 steps. Contents Page 1 Digitization and Predictive Planning ........................................ 177 2 Predictive Planning: What is it about? ........................................... 177 2.1 Significance for controlling in Medium-sized companies ................................ 177 2.2 The benefits of automated planning procedures and the maturity of medium-sized companies ... .................................................. .......................... 178 3 How do companies use and implement predictive planning? A practical guide ................................................ .............................. 179 3.1 Step 1: Identify the starting points ............ ........................... 179 3.2 Step 2: Data management - creating a solid basis ............ .. 180 3.3 Step 3: Single Point of Truth - centralize data storage 180 3.4 Step 4: Integration into the planning .......................... .................... 183 3.5 Step 5: Application in the planning - a "how-to" for the start 184 4 Conclusion: What are the limits of use? ................................. 186 5 References .............. .................................................. .......... 187 175 Predictive Planning in SMEs n The authors Dipl.-Kffr. Simone Doerfner, Chief Marketing & Communication Officer and member of the management of CP Corporate Planning AG in Hamburg. Dipl.-Inform. Matthias Kläsener, Chief Executive Officer of CP Corporate Planning AG in Hamburg. 176 Organization & IT 1 Digitization and Predictive Planning Advancing digitization is increasing the speed with which companies have to make their decisions. This increases the market dynamics and at the same time reduces planning security. Greater automation of control processes and real-time monitoring are important tasks of CFOs in order to control digitization as business partners. The greatest attention is paid to the forecast speed in order to enable the company to act quickly in its decisions. Factors for increasing the forecast speed can be, on the one hand, the focus on the essential control-relevant variables. Secondly, automated planning processes can significantly accelerate the creation of forecasts.1 In the following chapter, selected methods and approaches are presented how medium-sized companies can use automated planning processes and shows how widespread use is already. This is followed by a practical guide in which, on the basis of a BI solution, corporate planning implementation and application examples are shown.2 2 Predictive planning: What is it about? Predictive planning refers to methods and models that enable a view of future developments. This includes in particular scenario planning, simulation calculations, trend calculations and forecasting during the year. The method is based on the recognition of patterns in existing data stocks, through the analysis of which the system independently generates suggested values. 2.1 Significance for controlling in medium-sized companies The driver for predictive planning is called big data. Thanks to new technologies, corporate planners have a steadily growing amount of collected data at their disposal. With the advancement of technology and its increasing maturity, users who have no IT background knowledge can now also work with solutions. Modern technologies follow intuitive operating concepts and are very easy to access and implement as cloud solutions. This is of particular benefit to medium-sized companies that often do not have the resources for complex and time-consuming procedures. Modern solutions for predictive planning processes also offer users the option of merging data from different sources. 177 Predictive Planning in SMEs 1 Cf. Zillmann / Lünendonk, 2018, p. 10. 2 CP Corporate Planning AG is a German provider of controlling software with a focus on corporate management. The software solution contains over 300 business planning functions and enables medium-sized companies to digitally transform their corporate planning. save time in day-to-day business. Integrations between parallel systems not only ensure that information can be processed centrally. With the help of predictive planning, medium-sized companies gain faster and more reliable knowledge about their own company and the market environment.These new findings include, for example, recognizing relationships between different variables or patterns within the data.3 Companies can incorporate all of these findings into planning. 2.2 The benefits of automated planning processes and the degree of maturity of medium-sized companies According to the study "Predictive Planning and Forecasting takes corporate planning to the next level" (9/2018) by the Business Application Research Center (BARC), the relevance of predictive planning processes increases for 75% of the participants medium-sized companies in the DACH region. Above all, companies expect to use predictive planning methods to improve the planning processes themselves and to achieve meaningful results more quickly. However, only 22% of the companies surveyed use predictive planning methods in practice, while a full 59% even stated that they had not yet had any experience with them. However, at 86% most companies are currently either actively developing the necessary skills or planning to develop them in the future.4 The following 7 beneficial areas of application of predictive planning are seen by medium-sized companies as leading: • Higher quality and accuracy of planning and forecasting • Reduction of planning effort • Short-term and faster forecasts and prognoses • Proactive planning and control • Early detection and forecasting of events • Identify and evaluate cause-effect relationships • Relieve planners of routine tasks and focus on value-adding activities The aspect that predictive- 65% of the participants in the BARC study consider planning methods to offer a higher quality and accuracy of planning and prognoses as important. 59% expect the planning effort to be reduced. After all, almost half of the participants (49%) see short-term, faster forecasts and forecasts as an advantage, as well as proactive planning and control (44%). 42% expect the benefit of early detection and prognosis of events. Furthermore, for 40% the identification and evaluation 178 Organization & IT offers 3 Tischler / Fuchs / Engel, 2018, p. 10. 4 Tischler / Fuchs / Engel, 2018, p. 5ff. cause and effect relationships are of great benefit and 36% see that predictive planning methods relieve planners of routine tasks and instead concentrate on value-adding activities. 3 How do companies use and implement predictive planning? A practical guide Every company can improve its planning and forecasting using appropriate predictive planning methods. This requires a focused approach in order to use the available resources accurately and to get results quickly. The guideline in Fig. 1 shows controllers in concrete steps how they create the optimal conditions for using predictive planning methods in their company. Data management Integration in planning Identification of starting points Application in planning Single Point of Truth Step 2 Step 3 Step 4 Step 5 Step 1 Fig. 1: Prepare the use of predictive planning methods in 5 steps 3.1 Step 1: Identify the starting points First of all Controlling define those positions that offer potential for automated planning processes. Where is there potential for improvement? To do this, it is advisable to first collect the main drivers of the company. Which are the financial value drivers, which are the operational ones? For this purpose, all routine tasks should first be documented in the data collection so that controlling can then identify together with the planners which data is planned regularly and with recurring systematics, and where statistical processes could support. A pragmatic approach for working out the items is to first undertake the cost-side analysis and all items that are usually planned as part of the budgeting process on the basis of the previous year's values. Examples of this are current contracts, depreciation and standardized processes for provisions and accruals. Then all 179 predictive planning routine tasks that arise in medium-sized companies are to be collated in the data collection. These are precisely the starting points at which statistical methods can provide support. Even if it may seem marginal or self-evident: early communication with everyone involved makes a significant contribution to success. It is advisable to include all departments and colleagues whose support is required in the implementation process at this early stage. It must be clarified what information needs the individual planners have and then specific responsibilities should be assigned. Predictive planning approaches should offer the planner valid suggested values ​​as orientation aids and thus make a value-adding contribution to the efficiency of corporate planning. 3.2 Step 2: Data management - creating a solid basis In order to use the potential of modern planning methods, it is essential to work on the foundation and thus on the basis of all available data. The more precise, complete and error-free the underlying master data is, the higher the quality that can be generated by controlling for the automatically calculated plan values. The topic of data quality and data management should therefore be the first item on the controller's agenda for implementing predictive planning approaches. n The following individual tasks contribute to success: • Obtaining sovereignty over financial and non-financial master data in the company as an organizational consequence. This data also includes sales data - e.g. from CRM - as well as data from production or the HR area. • Analyze the existing data quality and check its consistency. • Eliminate data gaps and define the necessary master data for automation. • Check additional data sources and their integration. • Create a uniform data platform including the necessary interfaces. 3.3 Step 3: Single Point of Truth - centralize data storage The principle of so-called "platformization" is becoming increasingly important in the context of digitization. The aim of the company is to create a single point of truth in order to avoid data redundancies and to limit the scope for interpretation as far as possible. Every value should only exist once and it should be possible to trace back to its origin from every reference. In practice, this is often mapped through the use of suitable standard software, which, based on a uniform data warehouse, ensures horizontal and vertical 180 organization & IT data integration. This technological prerequisite also forms the basis for the entire automation potential that can be used in corporate management (see Fig. 2). Single point of truth information for all report recipients Technical integration at all levels One technological platform Fig. 2: The single point of truth - technological integration of corporate planning and central data storage In business practice, the creation of a single point of truth is essential - The concept is often accompanied by a data integration project. The processes of data export and data import are homogenized. This means that modern controlling systems have direct access to the company's actual database - exactly where it is collected. The technical integration of the controlling platforms with the upstream systems takes place via a data warehouse. It is thus possible for the controllers to trace a value down to the lowest level from the respective applications. In financial accounting, for example, this represents the document level. In CRM or ERP systems, it is the individual data record. Software providers usually already have the right interfaces to the respective upstream systems for these applications. The Lünendonk study 2018 "The market for business intelligence and business analytics in Germany" reflects the importance of this topic.5 Open interfaces and decentralized distribution of data and reports are the TOP topics that medium-sized user companies place on manufacturers as requirements. 181 Predictive Planning in SMEs 5 Cf. Zillmann / Lünendonk, 2018, p. 10. Fig. 3: Example of tracking a value (drill-down) with a controlling platform down to the document level in financial accounting 182 Organization & IT Die The advantages are as follows: • Avoidance of isolated solutions and errors due to heterogeneous import and export processes and sources; • Avoidance of black box behavior through transparent calculation paths in the value flow; • Avoidance of data redundancies in reporting; • Recognition of cause-effect relationships in the context of the deviation analyzes; • Creation of the prerequisites for company-wide real-time reporting. 3.4 Step 4: Integration in the planning In addition to the uniform technological basis that needs to be created, the individual sub-plans are integrated into a company-wide overall plan based on this (see Fig. 4). Intercompany Segments Legal Management… Balance Sheet Integrated Financial and Success Planning… Sales Cost Centers Personnel Investments Fig. 4: The integration of all operative sub-plans in the corporate planning and the continuous flow of values ​​up to the consolidation In this step the focus is on the development of a continuous planning model and the Linking the operative sub-plans with the integrated financial planning and, if necessary, a continuous flow of values ​​up to consolidation. The more intensively a planning is integrated, the more precise the planning results are usually 183 Predictive Planning in medium-sized companies, the more interlinked the different planning levels (success - financing - change in assets). The view of the success and financial situation based on the operative sub-plans becomes more meaningful as soon as the dependencies between the sub-plans have been taken into account and these have been seamlessly incorporated into the profit planning. This integration ensures that the planning model represents a suitable image of the company and its environment and is therefore suitable for supporting the best possible alignment of this company with its internal and external relationships.6 At the same time, the planning model represents the necessary basis for all planning procedures to be used. 3.5 Step 5: Application in planning - a “how-to” for the start Due to the increasing volatility, the topic of making planning more flexible is gaining in importance. A major challenge of controlling is to provide the management with meaningful budget figures with little use of resources and a fast turnaround time. Different aspects of the top-down orientation are the most common technical approaches today to make planning more efficient. The aim is to increase efficiency in the planning process through as few coordination rounds as possible. n The following approaches can contribute: • The distribution of centrally prepared values ​​for the individual planning areas; • the provision of planning parameters such as interest rates or raw material prices; • the definition of top-down guidelines. Stochastic and mathematical processes - which are part of the predictive planning approaches - are used to a large extent to create the forecast. A stochastic process is the mathematical description of temporally ordered, random processes. Thus, probability calculations in connection with mathematical estimation methods (mathematical statistics) form the mathematical sub-area of ​​stochastics. Stochastic processes and procedures are used in operational practice for forecast calculations and require existing data as the basis for the survey. Since the forecast as a planning scenario always contains the actual development and thus real series of numbers from the past, it is particularly suitable for updating planning procedures. When using a stochastic method, it is initially relevant for controllers to select the method that suits the company's business model. 184 Organization & IT 6 Cf. Tischler / Fuchs, 2016, p. 24. Fig. 5: Example of trend planning 185 Predictive planning in medium-sized companies n The following trend methods are often used in operational practice: • Linear trend (regression) • Logarithmic mixer Trend (regression) • Exponential trend (regression) • Geometric trend (regression) • Exponential smoothing with variable dynamics • 3-fold moving average with variable weighting • 5-fold moving average with variable weighting • Updated arithmetic average Suitable trend calculations for the respective business model can be selected using the approximation method. This means that there is no specific stochastic method to be used for a specific application. The following procedure can be used to check which method suits the company's planning model: 1. First, select values ​​that are suitable for the use of a stochastic method. As a rule, these are positions with a high potential for standardization and the longest possible past period. 2. The trend method can then be analyzed using the actual database. A visual representation is recommended to identify “outliers”. 3. Now the procedure is to be applied to a period oriented towards the past. The deviation that occurs shows where the mathematical procedure can still be adjusted. 4. Finally, the selected procedure must be adapted to the forecasting and applicable. When applying the corresponding procedures to the status analysis and the calculation of future planned and suggested values, standard software solutions in controlling often prove to be expedient. 4 Conclusion: What are the limits of use? In addition to accelerating and automating forecasting, the greatest potential for statistical prognoses is where one can carry out projections from leading indicators on a detailed level. Even if the extrapolation is based on an optimally prepared database, there are many influences that cannot be derived from data relating to the past. Plan models should therefore be reconsidered and revised on a rolling basis. Planning should be anchored in the company as an integral part of corporate management and the basis for all decisions. 186 Organization & IT, the responsibilities for fulfilling the planning activities and decision-making powers must be clearly regulated throughout the company. The central driver in dealing with volatility7 and digital change is an open performance culture. In many cases this means a fundamental change in culture. The greater planning uncertainty requires both trial and error procedures and a culture of open information exchange in order to exploit the potential of the possible planning procedures through digitization. Flexible working methods and dynamic teams are therefore essential for the use of predictive planning methods. Predictive planning has the potential to decisively further develop corporate planning. Professional software solutions ensure sustainable success and the efficiency of integrated corporate planning. This combines the strengths of controlling with the advantages of the machine: If creativity is required in practice, planners and controllers are responsible for these tasks. If, on the other hand, there are clear rules for extrapolation, it is potentially a task for the machine or the software solution. 5 References Schäffer / Weber, Digital Transformation - Controller superfluous? In Controlling & Management Review, special edition, 2017. Tischler / Fuchs / Engel, BARC study: Predictive Planning and Forecasting takes corporate planning to the next level, 2018. Tischler / Fuchs, BARC user study: Integrated corporate planning - Maturity level of German-speaking companies, 2016 Zillmann, Lünendonk whitepaper “The digital transformation of exam-related consulting”, 2018. 187 Predictive planning in medium-sized companies 7 See Schäffer / Weber, 2016, p. 12. 188 Organization & IT

Chapter Preview


Digitization in controlling, Predictive planning, Blockchain, Artificial intelligence



With increasing digitization, controlling is facing a radical change and has to face the digital challenges. This ranges from the automation and standardization of processes, through shortened periods of time for data acquisition and analysis in management reporting, to the improvement of planning and budgeting. These disruptive changes also require the controller to take on new roles and acquire new skills and ways of thinking. The authors show what effects digitization has on the work of the controller and on the controlling processes.They offer you instruments, techniques and best practice examples to make controlling fit for the future and to use the opportunities of digital transformation.


  • Digitization and the future tasks of the controller

  • Change of processes and role profiles in controlling

  • Effects of big data, predictive planning & analytics, artificial intelligence and blockchain on controlling

  • Control of digital business model innovations

  • Optimization of financial planning through driver models and scenario simulations


Digitization in controlling, Predictive planning, Blockchain, Artificial intelligence