That data is the fuel of the digital economy is a frequently heard statement. But what does that mean for the engines that are supposed to use this fuel, i.e. the systems that process data to move a company forward?
For example, the digital engine of a company cannot fill up the necessary fuel at a central gas station and, moreover, one and the same engine must be able to process different types of fuel, namely different types of structured and unstructured data.
The challenges this poses for companies and what the corresponding solution approaches look like can be illustrated using practical examples from e-commerce and the world of banking.
In e-commerce, more traditional data such as master data and data on ordering or accounting processes are supplemented, for example, by data that provide information about what arouses the attention of which customer at what time via which sales channel. Customers generate this additional data, which an online retailer should take into account if they want to be competitive. For example, it may be relevant whether a new customer came to the company’s homepage via a comparison platform and is therefore perhaps more price-conscious than other customers. If a customer also looks at the ratings of other customers on the homepage before making a purchase, this can be a sign of more quality awareness. If there is a point where customers frequently abandon the purchase process, appropriate improvement measures should be taken.
The world of finance
A look at the world of finance shows that a classic bank now has competitors on two fronts. First, it’s the big tech companies like Apple and Google that are revolutionizing certain services, such as payments. It is estimated that Apple Pay is activated on more than 500 million iPhones, and according to Google Pay, it now has 150 million users in 30 countries. Secondly, there are the neo-banks and fintechs that focus specifically on service modules and do not have to cover the entire spectrum of a universal bank. Right from the start, most competitors rely on cloud-based and modular IT landscapes. The flexible integration of third-party applications is planned directly. The classic banks must also upgrade their data in order to remain competitive. They need to install new IT systems that produce additional data and merge that data with that from the existing legacy systems. In addition, the entire systems must be protected against an increasing number of cases of fraud, which is particularly difficult in such a heterogeneous IT landscape.
Modern analytical platforms play a central role in overcoming the challenges described. Classic relational database systems, statistics and visualization programs are usually unable to perform the kind of big data processing that is required. New data storage and analysis systems must be used that work in parallel on up to hundreds or thousands of processors or servers. Modern analytical platforms enable companies to implement customer-oriented, digital processes combined with intelligent data analysis. Available data pots, which are often scattered throughout the organization, can be brought together and the analytical possibilities expanded.
Efficient analytical platforms should be able to be operated in the cloud in order to be used flexibly and be technically open so that, for example, proven and cost-effective open source solutions can also be used. For example, license fees can be saved by transferring statistical analysis software to free, open-source computer languages such as Python and R. The platform must be designed in such a way that the entire analytical value chain can be mapped. It must combine data storage, analysis and the presentation of results in such a way that no different resources are required for these steps, but a single user can manage them with the support of the system.
In addition, an analytical platform can and should also enable the use of machine learning (ML). The use of ML is not required by the sheer volume of data to be processed, but by the need to also include unstructured data in the analysis. For example, ML plays an important role in identity control in online commerce and online banking, especially in the form of advanced text and face recognition software. Appropriately trained ML algorithms can, for example, check the authenticity of the identity documents photographed by the applicant. ML algorithms for face recognition can also be used to compare the photo on the ID document with a recent selfie of the applicant and at the same time verify that it is a live selfie transmission of a real face.