Requirements for the analytics platform
One of the motivations for the project was HORNBACH’s goal of analysing reviews on social platforms such as Google Maps. These reviews are provided by customers after visiting one of HORNBACH’s hardware and DIY stores. They include such information as which store the customer visited, how they felt about their buying experience, etc. The aim of the analysis platform is to use anonymised data to determine how customers view their shopping experiences at HORNBACH stores. The company also wanted the analysis results to be presented visually and attractively, with the end goal of identifying potential for increasing sales. The platform also needed to be easily scalable in order to keep pace with increasing data volumes from additional use cases – while still remaining manageable from an operations standpoint.
Technical implementation of the analytics platform
The platform was created entirely using standard Microsoft Azure platform-as-a-service components. Data was integrated using Azure Data Factory via linked REST APIs and initially deposited to a landing zone in the Azure Data Lake Store. The raw data was then connected to the Azure SQL Data Warehouse using PolyBase. The data is processed and persisted in the database for a query-optimized data model. Sentiment analysis of unstructured text data is performed by Azure Cognitive Services, and services for translating foreign language entries and extracting key phrases are also used.
A semantic layer (SSAS tabular model) was added to the enriched data to form the reporting basis for analysing reviews in self-service mode. To ensure that the analyses are both intuitive and comprehensible, the KPIs were developed in close consultation with the specialist departments.
Data visualization is performed using Microsoft Power BI, another cloud-based software-as-a-service solution. Dashboards can be accessed through the Power BI portal as well as on mobile devices.
The project (and its streamlined interaction with Azure Data Services) was so successful that other Microsoft Azure data services have now been implemented at HORNBACH, the key ones being Azure Databricks for data science tasks and Power BI in the specialist departments. The interaction of the aforementioned components is illustrated by the diagram below.
Cloud BI brings benefits
The use of managed services in the Microsoft Azure Cloud, which dovetails perfectly with the existing project framework, has paid off. The comparatively easy entry into highly scalable services is particularly beneficial for users looking to access new technologies without having to develop their own operational know-how. Microsoft typically guarantees an availability of 99.99%.
The components are set up using agile processes and quickly deliver the first usable results. The cloud-based operations and potential for global scenarios relieve the burden on IT staff and provide flexibility for additional test environments. Other benefits for HORNBACH include faster innovation cycles, increased reliability, and reduced costs thanks to the synergy and efficiency effects provided by Azure’s public cloud concept.
Expansion of the analytics platform
The database is gradually being expanded to include additional social networks and demographic data. It will also allow systems located in the on-premises data centres to be easily connected via gateways. The components used can be scaled up with no loss of performance to handle increases in data volume and usage. For even more precise text analysis results, industry-specific models designed for DIY stores are being trained and used.
HORNBACH’s new analytics platform has thus prepared the company to face many of the challenges set to arise from the increasing digitalization of its business model.
- Microsoft Azure Data Lake
- Azure Cognitive Services
- Azure SQL & SQL Data Warehouse
- Azure Data Factory
- Azure Databricks
- Power BI