Business analytics from the Cloud

Current trends such as big data, the sharing economy and Internet of Things are heavily dependent on cloud-based applications. The benefits of cloud services range from cost savings on software licences to the targeted use of existing human resources in future-oriented developments of core businesses and immediate service sourcing without a lengthy procurement processes. Particularly for small and medium-sized companies or for companies with a low IT budget, cloud services are a very good alternative to internally operated applications and IT infrastructure. Easy access to professional services is possible without great effort or expense.

For example, applications can be used from a cloud environment any time they are needed, or updates and maintenance can be taken over directly by the service provider, thus relieving the workload for the internal IT department. Access to IT services is also simplified and can be very quickly adapted as required. In addition, pricing models are possible that only charge for the effective use of cloud services and are tailored to meet client needs. The Cloud can also have a positive impact on organisational business operations such as work arrangements. For example, collaboration across global teams and support for cross-functional processes are made much easier, as data and applications are centrally managed and accessible from anywhere, at any time. Thus no direct data exchange between team members is necessary.

Processes can be handled without media disruptions throughout global value chains and numerous companies. This benefit helps to build business ecosystems with stakeholder partnerships, which is an important feature of digitisation. Because business ecosystems require highly active global value chains.

Solutions are largely freely scalable and can be adapted to business developments at any time. This makes companies more agile and adaptable to external conditions and changes. Ideally, the use of cloud services also has a positive effect on the innovative strength of a company and the emergence of new business models. This due to the elimination of the often very high investments in IT infrastructure or software development required for classical IT projects as opposed to a cloud solution being applied.

Nevertheless, careful consideration is required to decide which services should be sourced from the Cloud, if any. SMEs should therefore first design a sourcing strategy that can then be put into practice. Based on business requirements, environmental influences and the current IT landscape, it is decided which services are to be provided internally and which ones are to be obtained from the Cloud. Obstacles to designing a sourcing strategy can result from a lack of transparency in the IT landscape or a lack of skills to evaluate appropriate cloud services and integrate them into one's own environment. In addition, there are often budget constraints that prevent long drawn-out evaluations.

Traditionally, companies analyse data to measure performance: Strategic goals are derived from a general corporate strategy, which are collected, for example, in a balanced scorecard (Kaplan & Norton, 1996). The achievement of the targets is then monitored using indicators known as key performance indicators (KPIs). The calculation of indicator values is usually based on the aggregation of data, for example the aggregation of sales revenue for the "sales" indicator, which is displayed in a dashboard. Further analyses are usually necessary when goals are not reached. If sales figures are too low, questions arise ("In which region", "For which products?, etc.), which can be answered using detailed tabular-graphical representations (reports) or interactive multidimensional analyses (Codd, Codd, & Salley, 1993).

Such analyses are always retrospective. They show which goals have not been achieved in the past and permit an understanding as to the reasons for this. This can help to avoid similar failures in the future. In many cases, however, failures are attributable to risks or missed opportunities due to frequently made decisions that are that are therefore known (for example, decisions on the granting of insurance policies). Chances include good timing and accurate predictions. For example, if a customer is able to make the right offer at the right time, such as while visiting a web shop, this has a positive effect on the success of the business. Correct predictions lead to the optimisation of process flows. Anticipating certain events, such as the imminent battery life of a forklift truck, leaves enough time to provide a replacement so that logistics processes are not interrupted or delayed. Risks include fraud, customer churn, non-payment of loans or serious insurance claims.

In all such cases decisions can be automated. The prerequisite for this is that enough data about past decisions, their context and consequences exist from which future decisions in the sense of predictive analytics can be derived. For example, customer reactions to past marketing campaigns can be used to better plan future campaigns or to address customers specifically and individually. Similarly, fraud cases can be prevented by analysing patterns of past fraud cases and recognising them in current cases. Many companies use corresponding analyses in various areas (Siegel, 2013) to proactively ensure that goals are reached instead of retrospectively analysing failures.

In addition to such operational decisions, however, the data also have strategic potential. When customers purchase and/or use products or services via digital channels, this leaves electronic traces from which knowledge can be extracted. The Internet of Things, for example, also generates usage data for products that are not consumed via digital channels. The extracted knowledge can be used, among other things, for product innovations or the development of new business models. Understanding the use of a product that deviates from the developers' expectations often makes it possible to surmise the underlying needs of the users and to derive new products or business models from these findings. The same applies to contributions from (potential) customers in social media ("customer intelligence") or to news about competitors ("competitive intelligence"), from which market knowledge can be gained that is essential for innovations or strategic decisions about market position.

How can the above-mentioned requirements be met, i.e. how can SMEs be provided with cost-effective but individually tailored and largely "maintenance-free" analytics solutions that they can use to fully exploit the potential of their data? The provision of a cloud-based analytics-as-a-service at least avoids the acquisition of hardware and outsources systems maintenance. However, this does not in itself meet the other requirements, especially the need for individually tailored solutions; further components are necessary here.

A current research project in cooperation with Informatec investigated how these could be designed. Based on the findings to date, our proposal for an analytics-as-a-service solution for SMEs consists of the following concepts:

Highly modularised BI software: To ensure that the individual needs of SMEs can be taken into account, they must be able to assemble their solution from a highly modularised set of components. This means that they can choose or define individual analyses to meet their information needs. Results of data analyses can be presented in the form of diagrams, tables or predictive models. For the individual definition of diagrams and tables, SMEs must be able to select which key figures (e.g.  sales) are to be displayed and according to which criteria they are to be grouped or filtered (e.g. sales by product, region, sales manager etc.).

Standardised and formalised description of BI components: To enable automated consulting at a later date (see below), a standardised and formalised (that is, machine-readable) description of the BI components must take place. This serves as a vocabulary, which in turn can be used to describe entire BI solutions.

In addition, a case base is built up. For each BI solution already implemented, the characteristics of the respective SME (e.g. industry or relevant business processes) are recorded. The components of the implemented solution are also described and stored using the description vocabulary already stated. The results of the case basis show the kinds of solutions each type of SMEs requires. It can also be seen which components are frequently used together.

(Partially) automated consulting: Highly modularised data results in a confusing range of possibilities that are mind-boggling for SME employees. Therefore, a (partially) automated consultation is recommended: SME representatives provide some information in an online questionnaire, for example on their company's industry and relevant business processes. Based on the knowledge from the case base, it is now possible to make suggestions for analyses that have been carried out in similar cases in the past. When SME representatives make a first selection, the knowledge about the interaction of solution components – which can also be gained from the case basis – can be used for further recommendations.

The modularisation and standardisation of BI solutions and the consulting automation enable solutions to be individually tailored to the needs of SMEs on the one hand, and cost-effectively implemented on the other.

However, the demand for the above-mentioned simplified data extraction and integration has not yet been met. Depending on the complexity of an SME system landscape, it can be very difficult to achieve this. Analytics-as-a-service providers can provide interfaces for common standard business software.

Ideally, it would be enough for SME representatives to name the modules of this standard software – which can also take place using an online questionnaire – in order to extract the data with the correct semantics from the source system into the cloud environment. However, manual intervention will usually be necessary.