The Challenge
A European Leader Supermarket chain decided to design and implement a Business Intelligence Strategy and applications in order to increase competitiveness and profitability.
The Solution. Align Knowledge Engineering to Business Objectives
Customer Centric Positioning
Consumers are the ultimate arbiters of enterprise ability to identify and predict market trends and to procure and distribute products and services that represent desired customer value, at the right price and through the right channels.
Firms must be aligned to consumers’ continually evolving needs and expectations of value.
As a result, the ability to innovate successfully to create customer-centric differentiation is critical to the overall success of the sector and increasingly decisive in the survival of individual enterprises.
In order to achieve a Customer-Centric framework, we created a Business Intelligence architectural plan that analyzes the interferences (input) of all external factors on customers and the consequences on their final purchase decision (output).
DATACTIF Business Intelligence Open System
We designed conceptual, logical and data models and the adequate data warehouse, after an in depth audit of business processes and aims, IT infrastructure, human resources availability and experience, transactional and other data quality, qualitative and quantitative researches as well as business scenarios that should be realized.
DATACTIF® Business Intelligence System was deployed in a multilevel applications schema, following Customer Centric strategy, using supervised and unsupervised learning algorithms and performing Clustering, Association Rules, Classification and Prediction tasks.
Knowledge visualization in accordance to human abilities is the most important step in data modeling.
DATACTIF® allows real time, direct, substantive assessment of enterprise corporate knowledge through visualization offered by and at all levels.
Customers Segmentation
The biggest problem with a supermarket data is the huge, continuously changing number of product codes (new products, seasonal products, promotions that makes a product appears with a different code, etc…) that makes any analysis a very difficult task. In the other hand using only categories of products make decision makers loose information details that only products offers.
We opted for a two layers approach, training SOM first with an intermediary categorization (categories of products plus brands) and finally with detailed products.
We succeed to make a state of the art Behavioral (based on purchase behavior) segmentation.
We used unsupervised learning and Self Organized Map that allows to find groups of clients (Clusters) that exhibit a certain degree of similarity in respect to a number of features that describe these objects (e.g. transactions of a client).
We opted for 25 clusters solution (5 X 5) with scientific but also business criteria as it is important for a retail company to have the less possible groups of customer in order to design cost effective business and marketing campaigns.
We obtained a behavioral segmentation with 25 distinctive clusters, as we see in the above picture.
Features extracted values allows us to examine each cluster separately, finding how and why it was formed as in Figure 1, where we see that Cluster 11 is made of families with young children, that prefer biological products.
By classifying clusters based on data such as : total sales, net profit, etc... we obtained the economical impact of each cluster on enterprise profitability.
Hyper Clusters
Based on features extracted values of each cluster and on clusters similitude’s analysis, we could define 5 Groups of Clusters, Hyper Clusters .
We created Hyper Clusters because we could easily obtain Behavioral, Benefit and Life Style Segmentation results unified in a way that allows to the enterprise to design cost efficient large scale business strategies (deals with suppliers, price reduction or in store promotions) and marketing campaigns (for each Hyper Cluster a distinctive communication).
Customers Segmentation History
Customers Segmentation observed through time, offers a macroscopic point of view on customers evolution in a social and economic context, measuring in same time the efficiency of the Enterprise's strategy.
Customer Segmentation History, allows comparison for the same clients between two time periods.
Suppliers Performance Evaluation
In a Customer Centric Strategy, supplier’s evaluation has to provide knowledge beyond market shares and profitability performances, taking into account suppliers brands and their marketing strategy, brands impact to customers and through this impact the result in the relation between the retailer and its customers.
An overall Supplier Evaluation Index was created based on brands, by categories of products as summary of partial indexes such as: Suppliers Gross Profit and Sales Evolution, Market Penetration, Customers Segments evaluation and Brands impact to Customer Segmentation.
