Advanced techniques for advanced and predictive analytics complement conventional BI analytics in an increasing number of companies. Growing experience is allowing companies to complete the phase of experimentation, and they are now faced with the challenge of how advanced analytics can be used productively as a valuable success factor in the company, and how this can be rolled out across the board.
Data scientists play a central role in the implementation of advanced analytics. However, the line of business is becoming increasingly more important. There are various reasons for this: On the one hand, more and more business intelligence tools offer their users advanced methods for analyzing data along with data processing functions. In addition, business users must engage increasingly with data scientists in order to identify use cases and understand the results of mathematical analyses. This requires users to have a basic understanding of methods of the most important analysis concepts and the ability to correctly interpret the output of the advanced analytics solutions. It is not unusual for key users to conduct data science projects in the line of business. The proximity to the processes to be optimized may play a critical role here.
Are the statistical and mathematical knowledge of data scientists and the use of special data mining tools required for all tasks? Or does current software with intuitive user interfaces, pre-built use cases, and pre-configured algorithms also enable users in the line of business to obtain the necessary skills and thus become a ‘citizen data scientist’? Can lines of business achieve the equivalent added value that data scientists and data labs can record? And will organizational obstacles during the implementation of projects be possibly reduced thanks to their proximity to operative processes?
In order to be able to provide answers to these questions, BARC conducted an additional user survey. With 210 respondents from the DACH region and a wide range of industries, the survey covers various company sizes and provides an objective insight into the current and planned usage of advanced and predictive analytics within companies. The survey can be downloaded here free of charge: https://barc.de/docs/research-note-advanced-predictive-analytics-survey-2017
Below is a summary of four of the most important findings from the survey.
1.1 Advanced analytics is regarded as an important issue – but implementation has stagnated
Advanced analytics is now considered as very important by 9 % of the survey’s respondents. This is twice as many compared with the previous year’s survey. 40 % of respondents currently regard advanced analytics as important, 43 % as less important, and 8 % as not important at all. 63 % think it is very important in the future and 31 % think it is important. However, the number of companies in which advanced analytics is used frequently or in isolated instances has stagnated – 5 % of the surveyed companies use advanced analytics frequently, 31 % in isolated cases. These results are consistent with BARC market observations: The need for information on the subject remains high, and an increasing number of companies are employing staff with the corresponding skills or are investing in the training of employees. Companies that are embedding advanced analytics into a systematic strategy still represent a minority. The majority of companies introduce advanced analytics by means of individual projects, or through information gained at conferences and from relevant literature. The willingness to invest in the subject has increased during the course of last year, but most companies are still cautious when it comes to drawing up extensive programs and investing in the appropriate tools.
Figure 1: “How relevant are advanced analytics and predictive analytics to your company today and in the future?” (n = 170)
1.2 Companies are predominantly prototyping advanced analytics use cases
More than 50 % of the surveyed companies currently using advanced analytics are prototyping use cases or evaluating prototypes at the moment. 22 % have completed this phase and are deploying advanced analytics on a regular basis for aiding the decision making process (15 %) or have automated important business processes (7 %). 18 % are currently in the use case identification phase. Interesting is the difference between companies where advanced analytics are driven by the line of business and companies where data scientists or the IT department are responsible for pursuing advanced analytics. Companies with a line of business structure are predominantly in the phase of identifying use cases and evaluating prototypes, while companies focusing on data scientist or IT department input are more likely to be in the prototyping phase. Line-of-business-led companies are also more likely to use advanced analytics to aid the decision making process (18 %) but not to automate processes (0 %). However, process automation represents a priority for data-lab- and IT-led companies (13 %). This suggests that the knowledge of professional data scientists as well as IT vis-à-vis the line of business also allows other use cases, or advanced analytics are being used by LoBs with different expectations.
Figure 2: “In which phase of implementation are you currently in?” according to LoB-led and IT-&-data-scientist-led companies (n = 74)
1.3 Key users in the line of business remain the largest group of users – management is the driving force for advanced analytics projects
Key users in the line of business formed the largest user group last year too. In 2017, 25 % of key users in the line of business conduct data science projects.
20% of the surveyed companies deploy data scientists, with only just over 50 % of the data scientists being part of a data lab. Data scientists have gained in importance over the previous year when it comes to the implementation of advanced analytics projects. After all, 13 % of respondents continue to use external service providers. Management is regarded as the driver or pioneer of advanced analytics, followed by the finance/controlling LoB, with the BI organization playing a critical role as pioneer in third place.
Figure 3: “Who conducts or will conduct advanced and predictive analytics in your company?” (n = 171)
1.4 LoBs are more likely to conduct advanced analytics using analytical apps and BI tools
Companies where advanced analytics is led by the LoBs are more likely to conduct such analytics as part of a BI environment or to use specific analytical applications. In contrast, IT- and data-lab-led companies rely considerably more on data-mining software. Excel/Power BI is the most frequently used tool (44 %) at LoB-led companies, followed by R (38 %), which can now be easily combined with the majority of popular BI tools. Special programming languages such as Scala and Python tend to be used more by IT- and data-lab-led companies. The percentage of specific data mining tools used such as SAS, IBM SPSS, and RapidMiner is rather low for LoB-led businesses. IT- and data-lab-led businesses use these tools twice as much in contrast.
These figures become clear against the backdrop that data scientists and data labs often rely on specific, additional software. As a result, the implementation of use cases focusing on process automation requires e.g., software with functions to support model deployment and model management. Open source programming languages such as R may offer comprehensive statistical libraries and be easy to integrate with BI tools; however, these have frequently disadvantages in the management, performance and operationalization of models.