More detailed insight into daily business transactions, generated by a deeper analysis of their characteristics, will help businesses to lower the costs of doing business, serve key segments better and seize transient opportunities, says customer behaviour prediction firm Xineoh founder and CEO Vian Chinner.

The Xineoh predictive analytics platform uses complex algorithms at various stages of analysis to derive insight from the characteristics of customers and, specifically, missed opportunities for greater business and sales. Understanding this will enable businesses to meet customer demands and identify untapped opportunities more easily and more cheaply, he explains.

The platform has been designed as a service model focused specifically on analysing and predicting customer behaviour, the analysis of which assists businesses in increasing the impact of their investments and spending.

However, by more effectively anticipating customer demands, a company can increase the financial benefit of its daily processes, such as stock purchases, which, in turn, reduces waste and enables the company to grow organically using internal resources.

Effectively serving a target market or market segment is a core principle of hyperscale international companies such as entertainment streaming multinational Netflix. The company analysed customer behaviour and determined that it could achieve customer satisfaction rates of 90% with only a quarter of the total movies, reducing its costs to store content and serve customers, highlights Chinner.

“Having a near-real-time view of its customers’ behaviours enables the company to anticipate their demands over the short term and this enables it to position and market ahead of time and seize a greater share of transient market opportunities than most of its competitors.”

Further, Xineoh is involved in South African use cases such as predicting purchasing behaviour and patterns for retailers, which often face low demand in some regions or for some products and high and unmet demand in other regions or for other products.

Xineoh focuses on the retail, ecommerce, financial services and media industries as its key markets, but its systems can be used in many industries, he adds.

“Smart business systems also increase in effectiveness over time as part of a virtuous cycle where more data strengthens learning algorithms that underpin better customer service, which, in turn, produces more data that is fed back to further hone data-driven business processes.”

Behaviour analysis and prediction are also contributing to changes in business cycles, where companies are starting to produce products to meet anticipated demand, not just existing demand.

“The decision to adopt smart information technology systems is typically born out of business strategy. “Such adoption of smart systems involves three phases, and we are normally contacted during the second phase for proof-of-concept and pilot projects within business silos.

“The third phase is the industrialisation of smart systems and reforming a company into a responsive, data-driven and customercentric operation. Additionally, modern cloud-based business services can scale almost indefinitely, and small companies can access high-technology business systems and services as easily as large enterprises.”

Chinner highlights that a key benefit of analytics systems is that they make visible the latent variables affecting transactions, which companies are typically unaware of. Xineoh starts by cleaning and arranging a client’s data to feed into the platform to produce insights and, based on this, predictions for clients.

“Data science usually revolves around determining the inputs and predicting the outputs. However, with the modern computing power and architecture of smart systems, this effect is multiplied through the combined use of many algorithms simultaneously that produce detailed portions of information that narrow the range of expected outcomes based on detailed data and making predictions more accurately.”

As companies become more data-driven, the accuracy of predictions becomes almost disconcerting. It is entirely possible that, with sufficient data, a machine learning system could predict one’s behaviour better than the individual could, with a commensurate change to business services and models, avers Chinner.