Meet Your AI Copilot fot Data Learn More

How a leading fast-food brand implemented Precision Marketing to maximize profit

Lena Pan
Technical Writer Lead
Jul. 19, 2022
Photo by Jesson Mata on Unsplash
Brief Introduction:
  • Customer: A large, global fast-food chain
  • Products: Kyligence Enterprise and Kyligence MDX
  • Applications: Order analysis and multi-dimensional analysis through unified semantic layer analytics
  • Industries: Retail, and accommodation and food service

A Leading Fast-Food Chain Winning China's Local Market


A global fast-food chain with more than 30,000 restaurants in more than 100 countries, and serving 68 million customers each day, had plans to expand in China, its second-largest and fastest-growing market. Already operating more than 3,000 restaurants in the country and employing more than 100,000 people, The Brand was looking for a way to maximize profit potential for both existing and new restaurant locations. The key to success for The Brand would be to identify opportunities for creating specialized promotions for specific regions and locations, and delivering those products with fast, consistent service–faster than it was already providing with its ready-to-eat menu and combination meals. How would The Brand do this? By leveraging the data it had accumulated from its fully digitized operations and turning that massive store of data into actionable intelligence.


Fast-Food Delivery Blossoms in China


Because of restrictions enacted in response to the COVID-19 pandemic, fast-food delivery orders became common for home-bound customers in China. As a result, industry reports show that the number of restaurants offering their menus through online delivery platforms in China spiked to new heights. Research firm Statista shows the market for China's online food delivery business in 2020 exceeded RMB 664.62 billion, and the number of active users reached 468 million.


Legacy Approaches Presented a Challenge to Online Order Analysis


Fast-food delivery had also become an important sales channel for The Brand, processing around 60,000 delivery orders per-day, and generating more than one million lines of detailed data. The Brand wanted to use that data to find a competitive edge and improve their customers’ experience. But any innovations would need to be based on precise business intelligence generated through timely and accurate data analysis.The Brand’s data scientists use Oracle Business Intelligence Enterprise Edition (OBIEE) + Impala to support self-service, multidimensional data analysis. With this solution, data analysts don’t have to worry about the complexity of their underlying data when building multidimensional models and performing ad hoc queries through a web interface.However, the growing volume of data–and a corresponding increased demand for analytics–means The Brand’s data scientists face new challenges. For example, The Brand wants to track its daily sales volume as well as raw materials and packaging costs. To get this information, The Brand’s data scientists need to read and analyze the delivery order data from the day before. But when using its existing technologies during peak query hours (9:00am to 10:00am), the analytics systems slowed to a crawl and often returned time-out errors. Even during non-peak times, when The Brand’s analysts attempted multidimensional queries, response times were protracted.Using traditional approaches and legacy systems, The Brand faced the following challenges:

  • Unsatisfactory query engine performance: Due to limited data engine resources, query timeout frequently occurred during peak hours. The Brand’s legacy solution only supported analysis with a few dimensions, and within a short time range. If data analysts wanted to analyze the data from a relatively long time range (more than one month), or with more dimensions/measures (more than seven to eight), the query would time out.
  • Analysis topics cannot be changed on the web UI: If the analysts wanted to do complicated KPI calculations or other advanced analysis, they had to download multiple multidimensional analysis results, and then perform secondary processing with Excel and other tools locally. They did try other technical solutions. For example, data layering in data warehouse, materialization of data with different aggregation granularities through ETL, etc., but these approaches only raised new problems:
    • Inconsistent metrics semantics: The naming of dimensions and metrics were inconsistent, and IT professionals needed to develop a different data model for each business scenario, which was not only time and resource-consuming, but it also affected the timeliness of analysis. That means IT professionals were constantly "reinventing the wheel."
    • Unable to support frequent changes in business dimensions: Whenever the business status of a restaurant changed, or whenever there was a change of regional managers, historical data had to be refreshed–requiring a labor-intensive process.

The Brand Turned to Kyligence Intelligent OLAP


To overcome these challenges, The Brand turned to Kyligence to integrate data from different sources and build a multidimensional model that includes many key business metrics. What’s more, the Kyligence unified semantic layer helps The Brand to unify business logic across the company. With Kyligence, The Brand was finally able to achieve self-service, multidimensional analysis of detailed data from hundreds-of-millions of historical orders. Now they can run analysis from dimensions like restaurants, products, and channels, and analysis efficiency is greatly improved. Some key advantages gained by The Brand by adopting Kyligence Intelligent OLAP:

  • Integrated data models: Kyligence helps to integrate data from different parties with existing models, and builds a multidimensional star model that covers key business information such as order channels, products, promotion policy, restaurant, and time, providing a more diversified analysis perspective.
  • Intelligent model building: The Brand’s IT professionals use Kyligence Enterprise to connect different data sources, and Kyligence also offers a no-code platform for model building, accelerating analytics delivery. Furthermore, the Kyligence AI-augmented engine detects patterns from the most frequently used business queries, automatically making recommendations to help model optimization. Kyligence also supports the automatic refresh of slowly changing dimensions, so there is no need to run a global update when there are data changes.
  • Unified business semantic and access control: Kyligence MDX unifies the definition of business dimensions and metrics across different models, creating a business-friendly unified semantic dataset. Kyligence also supports data access control–at the project-level, table-level, and cell-level, by users or user groups.
  • Direct Excel connection: Users can directly perform drag-and-drop analysis with Excel PivotTables instead of manually integrating multiple data sources. Kyligence also supports dimension slicing and query roll-up/push down, as well as on-the-fly aggregation based on PivotTable results. Kyligence offers seamless integration with Excel functions and macro operations.

The Kyligence solution architecture is as follows:


The Brand Achieved Precision Marketing with Kyligence Intelligent OLAP


Kyligence Intelligent OLAP + unified semantic layer analytics provides a unified query entry point and solves problems like inconsistent data processing, and logic and query timeout. With Kyligence, The Brand now gets timely information regarding the profitability of delivery orders, and at a lower TCO than with its legacy systems. That means The Brand can make faster, better-informed decisions, and tailor their marketing strategies based on more precise business intelligence.

  • From query timeout to second-level query response time: Kyligence helps The Brand achieve stable second-level query response times for data from different years, or data that consists of thousands of rows and dozens of columns. Now data analysts can do self-service, multidimensional analysis on the detailed data from the hundreds of millions of historical orders from dimensions like restaurants, products, and channels. Business analysis and decision-making efficiencies have been improved; now The Brand generates precise, timely decision intelligence and can better respond to market changes.
  • Lower TCO: Kyligence helps The Brand achieve better query performance and higher concurrency support with a lower TCO, including a YARN resource savings of more than 50%.
  • Digital refining operations: By enabling self-service, multidimensional analysis, Kyligence helps The Brand to run a precision marketing strategy, customizing elements like regional–or even restaurant-specific–combo-meal and single-product discount strategies.

According to The Brand’s data product team, with Kyligence multidimensional models and a unified semantic layer, their analysis experience is significantly improved, and they can better serve other teams with their analytics capabilities. With Kyligence Intelligent OLAP in place and delivering significant improvements, The Brand hopes to make even more gains by working with Kyligence in more areas, such as unified query routing and ad hoc analytics.

Want to know the differences of the Semantic Layers and the Metrics Store solution? Read this blog!