Increasingly, companies are making use of visual analytics as a standard part of their analytical toolkit. The main reason for this is that in terms of hardware, licensing models and user-friendliness, the technical options in this area have vastly improved in recent years. In auditing, however, visual analytics is currently only used to a very limited extent.
Data analytics describes a structured process for extracting information from raw data. In auditing, such analyses are generally programmed and executed in, for example, ACL, IDEA, SQL or Python. The results of the analyses are inspected separately and can be iteratively refined using further data analyses.
Visual analytics is a specific form of data analytics that uses dedicated software such as Tableau, Power BI or Qlik. With the help of visual representations, graphics, graphs, diagrams, progressions and maps, relationships between items or sets of data can be identified or examined. The aim is to obtain an overview of the available data such that the resulting findings can be correlated with existing information. It is also helpful for identifying slight deviations from or anomalies in existing processes or (non-)existent dependencies. Further analysis of the individual findings requires a high degree of flexibility and adaptability in both the visualisation software and the relevant expertise possessed by the analysts using it. In short, the audit team must have the skills required to create and interpret visual analyses.
Other applications for graphical visualisations are found in reporting, as well as business intelligence (BI) dashboards. The main difference lies in the separation of programming from consumption; essentially, the various forms of graphical representation are predetermined, i.e. cannot be changed by users. If users notice something, they can only analyse it further if the appropriate dashboards for additional detailed analysis are already available.
When a rules-based data analysis yields a large number of hits, visual analytics can be used to analyse the root causes and identify relationships and dependencies. The resulting insights can be used to refine the rules for another analytical iteration.
When applied to data validation, visual analytics can be used to check whether the available data is complete, unchanged and non-aggregated, and consequently suitable for further analysis. Because IT architectures differ depending on whether the business involved is a company, subsidiary or division, and also depending on its business model, such comparisons must be individually configured and adjusted. Similarly, visual analytics is very useful for rapidly identifying inaccurate data.
Analysts use selected presentation formats to identify relationships between the various organisation units in a company. When planning audits or reviewing completeness, they can see, for example, which purchasing and sales organisations were used, along with the relevant warehouses, companies, bookkeeping systems and bank accounts. This enables audit managers to establish whether all the organisation units involved (in a technical sense) were properly included in the individual audits following a given audit cycle.
Various elements are used in visual analytics:
Auditors must also fulfil additional requirements for documenting each step in the process. The various options for doing this have improved, but are often given different names depending on the software. Thus a particular program may allow auditors to create bookmarks, set up stories, add comments, create subpages, save filters or data selections, step back through multiple operations, or make use of a full range of interactive functions. Other options may include the ability to export graphics in JPEG format, or to print (non-interactive) PDF files.
The best audit results, as well as the most efficient working methods, are based on a combination of data analytics, visual analytics and reporting. SQL and Python are more than powerful enough for data modelling or running more complex analyses. Employees can use software that features reporting dashboards to access any data models that have already been created, allowing them to inspect recurring issues. Flexible visual analytics software can be used to clarify individual details.
Frankfurt School’s Certified Audit Data Scientist course covers examples of the latest technological advances and their applications, including risk-focused and rules-based data analysis procedures, Deep Learning, Business Process Mining and Visual Analytics. Robotics Process Automation for auditors shows various ways to automate and support the follow-up steps in a technical analysis.