Тема: DATA-DRIVEN DEVELOPMENT OF LOYALTY PROGRAM FOR RETAIL COMPANY

Nowadays, many supermarket chains with a similar assortment and approximately
the same prices are present on the market. Therefore, customers with such a wide variety can
afford to choose where to shop and in what quantity. Methods such as special offers and
loyalty programmes are generally used to attract customers.
Loyalty programmes are designed to build a long-term relationship with the customer
to highlight a particular supermarket chain among others and motivate customers to increase
the number of visits and the average check-up. The loyalty programme can help to:
- retain customers who bring the greatest profit,
- increase the average check-up and the frequency of purchases
- reduce the cost for regular customers
- establish constant communication with customers
- collect a high-quality database
- get valuable information about customer behaviour
- increase brand credibility and awareness
- monitor changes in consumer behaviour and effectively influence it
- optimise marketing expenses through personal offers.
Most well-known retailers have their loyalty programmes, generally similar to each
other; hence, it is necessary to improve the user experience by personalising the offers. To
implement a more personalised loyalty programme, collecting as much information as
possible about customers, their behaviour, and habits is vital. Unfortunately, the information
supermarkets can collect about the participants of loyalty programmes is often quite limited,
mainly about the number and the date of purchases. Nevertheless, based on such data, it is
possible to segment customers using RFM analysis that allows to successfully divide
customers into groups and, depending on the results, determine which groups of customers
can be stimulated and how.
This research was conducted for a Russian regional supermarket chain that also has a loyalty
programme, which, however, does not correspond to current trends in this industry and does
not show high efficiency. The research has a practical value offering recommendations on
improving the existing loyalty programme.
The main research goal of this project is to analyse customer data for the period from
2020 to 2021 provided by the Company and perform segmentation of consumers into
9
different groups. This allows to visualise customers’ behaviour and to improve the existing
loyalty programme strengthening the Company’s position in the competitive market.
To achieve the goal, the following objectives are defined:
- Data preparation for further analysis
- Conducting customer segmentation and identifying key changes in customer
behaviour
- Data visualisation of results
- Problem identification and loyalty programme development.
In pursuing these tasks, various tools of data analysis were employed, including a
JupyterHub environment with Python libraries, Microsoft Excel and business intelligence
software (Microsoft Power BI). Work with data is organised based on CRISP-DM
methodology. The present work includes both theoretical aspects of studying the
development of loyalty programmes and the CRISP-DM methodology for data processing
and analysis. The research methodology is shown in Figure 1.
As a result of these stages, it is planned to obtain the following results:
- Determination of the number of loyal consumers
- Dynamics of changes in customer segmentation
- Identification of customers who may fall into the outflow.
The second chapter describes the theoretical foundations and features of developing
loyalty programmes, provides various types of loyalty programmes and justifies the choice
of a multi-level bonus programme as the basis of recommendations. It also explains the
chosen method of RFM analysis for further segmentation and describes the company
providing the data.
Theoretical
background
Method
selection
Data
description Data filtering
Modeling Conclusions Recommendations
10
In the third chapter, the selected method of RFM analysis is considered in detail, data
are described and prepared, and criteria for the application of the method are developed.
The fourth chapter examines visualised results for each city, where the analysed
supermarket chain is located. Furthermore, based on the results of segmentation, a new
loyalty programme is proposed with the possible results of its implementation described.
The study is concluded with a summary of findings and results with
recommendations for future work.
Купить

The main goal was to improve the existing loyalty programme through customer
segmentation to achieve the goal, the following tasks should be completed: data cleaning
and preparation, conducting customer segmentation and identifying key changes, data
visualization, problem identification and loyalty program development.
In the process of research, the following theoretical and practical results were
obtained:
1. Now, retail companies mainly use the loyalty bonus program since it is more
profitable for the company.
2. To determine outliers in data on customer purchase amounts, it is rational to use
the interquartile method, as well as a box plot for visualisation.
3. For all cities except the regional centre, the frequency index decreased the most.
In the regional centre, most indicators have mostly not changed.
4. In the regional centre, despite the high competition from other supermarkets, the
situation is the most favourable, since the number of new users exceeded the number of those
who left.
5. In general, the main problem throughout the network is the reduction in the
frequency of visits.
6. To solve the problem of the frequency of visits, as well as to increase the average
amount of purchases, a new multi-level loyalty bonus program was proposed with a
dependence of the percentage of bonus accrual on the number of purchases for the previous
month.
Overall, the research goal was achieved - all tasks and objectives were completed
successfully.
Recommendations for future work
The research of the characteristics of customer behaviour can be improved and deepened
with the help of a forecast of customer behavior for the next year by months, in particular,
the forecast of customers who may be in the outflow group. Also, further work involves
forecasting, taking into account the proposed recommendations and comparing these
indicators. In general, subsequent research should be directed to the practical implementation
and verification of proposals.
Купить