Being smarter with statistics has always been a business advantage.

Historically, it came under many different labels. First, with operations research, econometrics and business intelligence (BI), it was all about making better decisions. Then, in the upcoming e-commerce domain, it was primarily about cost and enabling transactions. Successful Internet companies adapted this to the full extent to create business value. With data mining, a differentiation potential appeared around customer convenience, such as providing relevant recommendations (‘If you bought this, then you may also like this’). With (language-independent) behavioural data, the results of analytics could be brought to a whole new level. Revealed preference theory, self-selection and transaction cost analysis are the economic basis for the introduction of behaviour-based recommendation services. The mechanism design of those services decides on the success and acceptance with customers. The ‘cold start problem’ plays a critical role in service initiation: predictive analytics needs a certain amount of data for significant statistics to create initial results. 

Suddenly, all types of data, including unstructured data, seemed valuable – not only to reduce the cold start problem. ‘Big data’ was born.

Nevertheless, data themselves do not make any business successful if they stay in a data tomb. The signal-to-noise ratio might be very low in big data. All data efforts can only create business value through advanced analytics. Corporations run analyses on massive data sets to find the signals in the noisy data.

Pattern detection via clustering, segmentation, (linear) regression, decision trees or data analysis with a focus on graph theory are the approaches to finding valuable business insights, requiring only ingestible data formats and no predefined business processes and performance indicators. Extracting knowledge from information sources – this is ‘data science ‘.

But still, knowledge extraction is only part of the game. How can you improve your business by means of this knowledge? How do you translate analytics into competitive advantage? How do you obtain management support? How do you set up a best practice data analytics project?

Storytelling is the art of reducing complexity of results so as to target a management audience. 

Advanced visualisation is crucial to transform complex data structures into comprehensible patterns for operational decisions. Business analytics is multi-disciplinary: the value can only be created inside an organisation with management commitment from a variety of departments (e.g., analytics team, IT, operational execution team) and suitable strategic planning.

What is important is to define the value in order to convince stakeholders of the potential for success. 

Quantify it for communication. People prefer things they can understand. 

Be an evangelist: the message must be suitable for the audience – and not based on the most impressive new statistical method. For business analytics projects, take care of the analytical, the process, the personal and the strategic perspective at the same time. Deliver value – the best sustainable competitive advantage – and prove it by comprehensible measurements. Take care that your CRM campaign management tool is actually able to provide support.

At first, data management platforms based on Hadoop as well as big data appliances for massive parallel processing were used for more traditional BI-like approaches to marketing. Their more efficient processing capabilities allowed for quicker results than previously possible. Nowadays, use cases beyond the traditional come into focus. In which areas should you look for a competitive advantage achievable by data analytics? What are examples of promising use cases?

Online retailers as well as financial institutions have long desired a 360-degree customer view. While this term in the BI world includes all relevant positions and, in the CRM, world also includes all communication contacts with sales, a true 360-degree view should additionally include all visited pages, where and when the customer lingered, as well as the leaving points. If available, that can even be enhanced by social network or browsing histories captured via advertisement tracking. This mixture of structured and unstructured data – if put to good use with predictive analytics – provides a detailed view of what the customer bought, looked for, needs and will most likely buy next. Propensity models, like the propensity to buy, are at the very centre of predictive analytics models for marketing, allowing more profitable relations with customers to develop.

Smart TVs, smart watches, cars, sensors in general, information security and telecommunications are only the first devices in a growing number of IoT-enabled smart gadgetry. They are sitting in the customer’s environment and ‘phone home’ with valuable usage and health information. Only big data technologies are capable of analysing these data sources for new service introductions and sales improvements. 

Whoever puts these data to good use will be the first to have competitive advantage over its competitors. 

Examples such as risk assessment for insurance companies and lead generation for cross and upselling any household, technology, over-the-counter medicinal products or grocery goods easily come to mind.

Despite the one-data-warehouse paradigm, all big corporations continue to run on multiple legacy applications with distributed data. This often hinders the optimisation of business processes. Big data technologies like Hadoop now allow, really for the first time, the gathering and extraction of heterogeneous data from multiple sources in an efficient way. This allows for service improvements, such as moving from next-day to same-day balancing in financial institutions as well as for real-time customer views including all contacts over all channels and event-based, immediately reacting client-facing systems.

Smart meters are being rolled out to homes everywhere. Vast volumes of data are starting to flow to electricity suppliers via smart grids. Most are ill-prepared to turn this into insights and actions.

Customer segmentation by energy patterns, for revenue increase via upselling, is one way of putting these data to good use. Another is the prevention of customer churn in deregulated markets; this is becoming more challenging with more competitors offering time-limited discount deals. Electricity demand fluctuates with the weather due to human behavioural changes; electricity supply fluctuates according to wind and solar power sources. For the suppliers, the reduction of direct energy costs via more accurate load forecasts matching demand and supply, as well as more effective tools for finding frauds and leaks, can bring a competitive advantage. Customers can profit by accurate up-to-date information on their energy consumption, and Internet-connected machines may be able to react on offers automatically.

The retail industry has been a first mover when it comes to adopting predictive analytics. Recommender systems are now standard features in every e-commerce shop. Real-time analytics for direct interaction with unknown online visitors close the gap with the digital footprints of identified repeat buyers. Ideally, retailers want to be able to predict consumers’ propensity to buy a product before it is stocked or, better still, before the production order is even placed. 

Imagine, trend analysis shows what market dynamics modify demand curves before they occur: the operational impact and risk mitigation in terms of supply chain efficiency, inventory cost control and intelligent distribution make every analytics improvement immediately worthwhile.