ES ▾

Machine Learning: Sounds Great, Used Scarcely

Joint study by Mains Lab, NAFI

Machine learning is already in use in public management, retail and sales, industry and power engineering, while medical and pharmaceutical companies are only about to start embracing it. The NAFI analytical center and Mains Group, a company specializing in education and information technologies, are presenting results of a study into the demand for machine learning technologies among big Russian businesses.
Automation of business processes is spreading widely across big businesses (80%), and the automation level will continue to grow. However, machine learning technologies are only used by a third of enterprises (33%).
publication date:
November 26, 2019
Machine learning
Machine learning

Main conclusions







There is awareness of machine learning technologies, nevertheless, they are not used at the majority of big businesses

CIOs expect further growth in business process automation, in particular, thanks to large-scale implementation of machine learning

As of today, machine learning is most frequently used in sales

The most promising areas for future machine learning implementation are operational excellence, analytics, and research

Machine learning introduction is primarily driven by cost reduction, while productivity gains and optimization of interaction among units are also among key drivers

The most common hurdles are the need for process reengineering and unpreparedness of company personnel


Process automation has become widespread among big businesses

For the purposes of this study, process automation means putting routine procedures or standard tasks under control of a digital information system. Examples of process automation are customer relations management (CRM) systems, automatic management systems for warehouses, supplies, manufacture, accounting, HR, etc.

Machine learning means a process in which a system (AI) processes a big number of examples (cases), detects patterns, and uses them to analyze and forecast new data specifications. For instance, a system can analyze complaints filed with a company, and make complaint forecasts.
decision-makers in IT at big businesses see machine learning as a useful tool that will become more in demand in the future



This creates a kind of hurdle for entry and active use of the technology. People are simply not ready to commit to a technology they do not know.
Top managers of the majority of big business are generally aware or have heard something of machine learning.
Most of the big businesses (80%) whose representatives were surveyed use automation in business processes, and every fifth business (20%) boasts automation of all business processes. The number of automated business processes at enterprises will continue to grow.
Even though business process automation is used by the majority of companies that participated in the study, only a third of them use machine learning (33%).
A major reason behind slow technology introduction is its delayed effect. Taking into account that AI based models need time to learn, and then a relatively long period is required for introduction and alignment, it may take a while until any economic effect is generated. In companies with a short planning horizons, the management just cannot afford waiting. That is why decisions are often made in favor of less efficient, but quick solutions, to the extent of hiring a big number of low-skilled personnel.
"Russian companies are becoming more prepared for digital economy. This is exemplified by our own products for big business automation based on machine learning technology. At the same time, the gap in up-to-date technology penetration in big and medium businesses may be massive. The study established that public management, finance, industry, retail, and healthcare were the leaders in the use of machine learning technology. Other sectors are lagging behind partly due to the nature of digitalization hurdles. Key problems we are facing in cooperation with the big business include processing of unstructured data, insufficient budgets, and shortage of qualified personnel. Based on the obtained data, we assume that we will enter a period of active transition to machine learning, data classification, and experience gaining in the next 5-7 years"
CEO of Mains Group
Sergey Khudyakov