29 ноября 2021
Article

How Do You Manage Data-Driven Changes?

Management Analytics as a Tool for Businessmen and Functional Directors
How Do You Manage Data-Driven Changes?
Content source: SKOLKOVO ММА

What do the tech company Google, retail platform Amazon, entertainment provider Netflix, and taxi app Uber all have in common, other than their dominant positions in the global business arena? They all use data-driven analytics to better understand their customers, industry trends, competitors, and, of course, their own business. Whether you are aware of it or not, data has become very much integrated into our daily lives, and its analysis has long ceased to be a prerogative of the giants with their impressive resources, having now become an important tool in the hands of a skilled manager, be they the CTO of a major communications provider with millions of users, or the owner of a family-run bakers at the end of the street.

Together with lecturers on the SKOLKOVO Masters in Management Analytics course, experts on project management and data analysis, we explain why you cannot afford to avoid data when making managerial decisions.

Five Reasons Why Data-Driven Management Works

With data analysis, companies can predict trends and even create them. Analysing consumer behaviour, browsing history and social activity enables you to obtain vital information about market trends, to quickly spot and exploit opportunities, and avoid guesswork when making decisions. The story of Netflix is a perfect illustration of predictive analytics in action. Analysing more than 30m views a day, 4m ratings and 3m searches on its platform, the company identifies the content that will be in demand by users. It was precisely down to data analysis that the hit series House of Cards came about.

Companies which work directly with a large number of users (telecoms providers, financial services and retail companies) have large volumes of data at their disposal, which means great opportunities for analysis and gaining insights. A certain major Russian mobile operator has around 50m subscribers, each of which generates myriad data segments – calls, geolocation, internet traffic, payments for communications and additional services, payments in retail outlets and many others - petabytes of information.

Dmitry Dorofeev
Former Vice-President for Strategy and Customer Experience at VimpelCom, Advisor on Business Strategy and Transformation, investor

There are industries and areas that depend the most on data, such as IT, banks, telecommunications and retail. Fighting to compete in such fields is now impossible without the use of advanced analytical tools. A small business which knows its customers by sight can get by without automated systems, but it still has to analyse its data in order, for example, to see which product sells the best, and to adjust its offers.

Data analysis is a broad concept. Along with machine learning (ML) models and recommendation systems for optimisation at tech companies or in complex manufacturing, it includes so-called “manual” analytics, that can also inhabit Excel spreadsheets. In almost every area where we are dealing with some kind of repeated actions, there is great potential for analytics. It may sound surprising but even athletes and those in creative industries, such as designers and media artists, rely on data. Personal recommendations on nutrition, training and regimes for athletes, and trend analysis and forecasting the popularity of cuts and fabrics for the next season for designers.

Emeli Dral
Expert in Machine Learning and Data Analytics, Co-Founder and CTO at Evidently AI

Data identifies where customer experience can be improved. In 2019, the fast-food chain McDonald's acquired the Israeli startup Dynamic Yield for $300m, a personalisation platform based on AI technology. This technology will allow the franchisee of the most popular restaurant in the world to display food depending on such factors as the time of day, purchases by other customers, and restaurant traffic. The company hopes in this way to improve its customer experience. Data analysis helps to offer customers what they need.

The modern business is moving away from global standardisation towards individual customisation. The customer does not want a standard product anymore, he wants a product that he personally needs with the range of characteristics, and at the time and place, that suit him. In trying to satisfy an individual request, companies collect more and more data, learn to find new correlations, generate, prove and reject thousands of hypotheses in the search for new opportunities to improve customer experience, including through better personalisation algorithms.

Dmitry Dorofeev
Former Vice-President for Strategy and Customer Experience at VimpelCom, Advisor on Business Strategy and Transformation, investor

Data helps find new opportunities to grow and save money. Using real-time monitoring of expenses and sales over several months or years, a company can identify various patterns, find hidden areas of growth or issues, and change tactics there and then. In 2018, digital marketing expenses for Coca-Cola amounted to $283bn, and by 2023 they would exceed $500bn. To ensure the huge budget for digital marketing was spent more efficiently, the giant began to use technology in image recognition on social media. Accurate identification of customers and their targeting, selection of the most advantageous channels and strategies increased the efficiency of tailored ads placed there many times over.  

The vast majority of marketing communications are done using analytics. These are recommendations for contact times, channels and personalisation of offers, and much more. By using the available tools to forecast demand, a company may find out unexpectedly that the promotional events in retail outlets have more of an effect on demand than product properties. Thus, using data analysis companies can not only predict demand for individual products or groups of them, but also manage it by way of promotional activities. Data analysis is a huge boost to understanding what is going on in your business and a huge boost to opportunities to manage it properly.

Emeli Dral
Expert in Machine Learning and Data Analytics, Co-Founder and CTO at Evidently AI

Data can be monetised. Google Analytics, Yandex.Metrics are the most obvious examples of projects successfully selling clients accumulated data and convenient tools for their analysis. However, the potential for creating value out of data is still very great, especially in areas that historically rely on huge arrays of information about users. New products and services that monetise data are constantly appearing on the market. So, for example, telecoms operators have learned how to make money by selling impersonal geoanalytics to retail chains, and to banks, data on the ability to pay of various customer profiles.

Finally, data analysis allows you to make better decisions in terms of business aims, quicker.  All businesses develop in a competitive environment, and top managers and business owners responsible for business results want to use all available opportunities and tools to gain the necessary advantages.

There are no decisions in modern business that can be made just by tossing a coin. Once, in mobile communications there was the simple relationship between a person and a SIM card: everyone had one SIM card, which is why the information systems of traditional operators are geared towards analysing  SIM cards, not subscribers. But the world has changed a lot. These days, a user can have a couple of phones, a tablet and multiple different digital services that he consumes even without a SIM card. To try to understand the user, and not the SIM card, requires a completely different order of intellectual efforts and tools for working with data. If some 15 years ago it was really not something to worry about, now that is completely impossible. If the operator continues to use Excel and count SIM cards, it will lose the battle with its competitors. Intuitive management decisions won’t be right, as the industry is far too complicated.

Dmitry Dorofeev
Former Vice-President for Strategy and Customer Experience at VimpelCom, Advisor on Business Strategy and Transformation, investor

Entrepreneurship is an activity that takes place against a background of uncertainty. Incompleteness and inaccuracy of data is not the exception but the rule.

If we knew for sure how every action will affect the result, the management process would be as straight-forward as you like. But this is not the case, and most managers have become used to living with a situation where decisions have to be made with a distinct lack of information. Data analysis is a sure way to reduce uncertainty, to better understand what is going on with the company, as well as an opportunity to compare yourself with competitors and global trends.

Emeli Dral
Expert in Machine Learning and Data Analytics, Co-Founder and CTO at Evidently AI

Management Analytics as a Tool in the Hands of the Manager

To use the opportunities afforded by data analysis and visualisation, the manager or business owner must know how to formulate tasks, what the results might be, what logic is being used to construct these results, how reliable they are and how they can be used in the decision-making process. Even if you have a team of professional Data Scientists at your disposal, there are certain tasks that must not be delegated.

I often see situations where data analytics is being used in projects just because it is technically possible. We can make a prediction – let's do it, we can construct recommendation models – let's construct them. But in reality, it becomes clear that results that are poorly applicable to the current business process are simply not necessary or that the wrong forecast horizon has been chosen. Data analysis is a tool to be applied with the right effort in the right place, and the person responsible for this should be the one who in understands in the first place the aims of the business and the reasons for them using this tool.

Emeli Dral
Expert in Machine Learning and Data Analytics, Co-Founder and CTO at Evidently AI

In the existing body of data, the number of signals it is possible to receive is infinite. To focus the work of the team of Data Scientists on the area most likely to prove useful to your business goal, the manager as the consumer of the data analysis results must know what they want to obtain, what resources are required for this, and be capable of properly formulating the task for the team.

One of the most important objectives of the telecoms company in customer experience is to build a network with a level of quality that best satisfies the customer’s needs. In practice, subscribers are dissatisfied not with an abstract quality of the network, but with it not working at points in their client behaviour. If you live on the outskirts of the city but work in the centre, then you want to have internet and a signal not just at home but also on the way into Moscow, and in the office and also at the shopping centre where you go with the family every week.

The traditional network planning model divides all investment into two segments – coverage in new areas and increasing the density in existing ones. The algorithm for calculating the location of new base stations is based on an assessment of the traffic of existing ones: the density of stations increases, and vice versa. Of course, the algorithm doesn’t take account of the movement of individual subscribers, otherwise the stations would be unprofitable.

It is easy to understand this from the point of view of a single customer especially if you yourself are the top manager of the company. You can go and bombard the Technical Director with complaints that in precisely your area there is no signal. When we’re talking about a large number of subscribers with different consumer habits and behaviour, this approach is unacceptable.

The hypothesis we tested was that constructing a network of base stations can also be done drawing on data on the movements of subscribers around a region. So that the scope of the search for insights is limited and not infinite, it is important to correctly formulate the task for the analysts. So, based on the data we have on the routes used by subscribers to move around the Moscow Region, we identified the priority clusters which require strengthening in terms of network quality.

Today, telecoms operators are taking steps to move away from the traditional model towards a new one. But, in order to improve the quality of data, there is still a huge amount of work to be done.

Dmitry Dorofeev
Former Vice-President for Strategy and Customer Experience at VimpelCom, Advisor on Business Strategy and Transformation, investor

The ability to independently interpret the results of data analysis and make tactical decisions can turn out to be critical in certain situations.

I participated in the development of a project for personalising marketing communications for a large bank. The idea was to make personal marketing product recommendations, to increase conversions and clicks from emails to the page featuring the offer. The team applied a lot of good machine learning models. However, results of the AB test turned out to be dispiriting. The new recommendations didn’t yield any significant growth. The Project Manager decided to analyse the process from beginning to end. It turned out that the recommendations were too low down, and not placed on the main screen, so the customers just didn't see them. Rather than continuing to optimise what wasn’t needed, namely the machine learning algorithm, adding more data and complicating the architecture, the manager made a non-obvious managerial decision – to change the design. Such skills are acquired through practice and a high level of awareness.

Emeli Dral
Expert in Machine Learning and Data Analytics, Co-Founder and CTO at Evidently AI

How to Become a Data-Driven Manager

In a competitive business environment, the demand for knowledge in management analytics is very high. Requirements for format and content on the part of business owners and top managers differ from those of technical specialists. To learn how to resolve tactical and strategic management tasks that are almost always characterised by some level of uncertainty, a close exchange of expertise, work on one’s own projects and analysis of market cases are necessary. The SKOLKOVO Masters in Management Analytics programme combines three educational blocks: strategic, instrumental and functional. Training takes place in online and offline formats, which facilitates networking and the exchange of experience.

Russian business today is the most active consumer of data. I will be moderating the Strategy and Change Management modules together with Professor David Saunders to bring expertise and the specifics of a Russian major telecoms business into the subtle science of working with data.

Dmitry Dorofeev
Former Vice-President for Strategy and Customer Experience at VimpelCom, Advisor on Business Strategy and Transformation, investor

The SKOLKOVO MMA is management training with data analysis tools. The programme comprises diverse disciplines. Students may possess knowledge about mathematics, statistics, analytics, or have strong applied knowledge in business. So that mastering the tools of data analytics and business disciplines is equally comfortable  for everyone, a “knowledge library” has been established, where all the basic required courses have been collected. This information will help structure existing knowledge and serve as preparation for the programme. 

My course on Machine Learning and Artificial Intelligence will suit managers who manage projects directly but who have a technological background. We will be immersing ourselves in the mathematical details of optimisation, trying various techniques and technologies to understand how we can affect the optimisation of specific business metrics using solutions to mathematical problems. The second course - on AI – caters to top management. As well as machine learning, we will examine different types of analytics: descriptive analysis of historical data, diagnostic analytics, predictive models both with and without ML, and prescriptive analytics. We analyse a large number of cases of other companies, engage practising heads of Data Science departments and other technical project managers who can share their successful and unsuccessful cases and work practices.

Emeli Dral
Expert in Machine Learning and Data Analytics, Co-Founder and CTO at Evidently AI

The programme’s lecturers include international experts with applied practical experience in data-driven business transformation. Graduates receive a Diploma of Further Vocational Education and the qualification Masters in Management Analytics, as well as a ready-made project for transforming their own company or business function.

(0)
(0)

Мы используем файлы cookie чтобы сделать сайт еще удобнее для Вас. Оставаясь с нами, вы соглашаетесь на обработку файлов cookie