Download PDF
How HelloFresh reduced promo abuse by 95% with Digital Trust & Safety
Technology Category
- Analytics & Modeling - Machine Learning
Applicable Industries
- Food & Beverage
Applicable Functions
- Business Operation
- Sales & Marketing
Use Cases
- Fraud Detection
Services
- Data Science Services
The Challenge
HelloFresh, the world’s leading meal kit company, faced a significant challenge with users exploiting their promotional offers, which was hurting their bottom line. The company initially tried to tackle these challenges internally through manual review processes in spreadsheets, but quickly found that they didn’t have the breadth of data they needed to effectively detect which customers were exploiting their system. The team decided it was crucial to seek out a more effective and efficient solution on the market instead of building their own capabilities. They were looking for a flexible model that could adapt to each of their market’s unique needs, responsive and knowledgeable customer support, and an adjustable pricing model.
About The Customer
HelloFresh is the world’s leading meal kit company, providing more than 600 million meals to 5.3 million households worldwide, in 14 countries and across 3 continents, in 2020. The company provides households with everything they need to prepare quality, delicious, home-cooked meals that require no planning, shopping, or hassle. Customers are supplied with every ingredient needed for thousands of HelloFresh recipes—all planned, sourced, and delivered to customers’ doorsteps. HelloFresh is dedicated to changing the way people eat, helping customers save money, access high-quality meals, reduce food waste, and take the stress out of meal time.
The Solution
HelloFresh chose to partner with Sift over the competition due to Sift’s unrivaled machine learning customization, the ability to increase or reduce thresholds depending on business needs, and hearing first-hand from other Sift customers how helpful the customer support was for them. The implementation process was quick and simple, taking only a week and requiring minimal code. Sift helped HelloFresh to confirm whether two people with the same address are part of the same household and trying to take advantage of promo offers—and block or unblock accordingly. Together, HelloFresh and Sift deployed 12 custom machine learning models to accommodate each market’s needs. With these customizable models, HelloFresh can easily personalize the solution to their markets.
Operational Impact
Quantitative Benefit
Related Case Studies.
Case Study
The Kellogg Company
Kellogg keeps a close eye on its trade spend, analyzing large volumes of data and running complex simulations to predict which promotional activities will be the most effective. Kellogg needed to decrease the trade spend but its traditional relational database on premises could not keep up with the pace of demand.
Case Study
HEINEKEN Uses the Cloud to Reach 10.5 Million Consumers
For 2012 campaign, the Bond promotion, it planned to launch the campaign at the same time everywhere on the planet. That created unprecedented challenges for HEINEKEN—nowhere more so than in its technology operation. The primary digital content for the campaign was a 100-megabyte movie that had to play flawlessly for millions of viewers worldwide. After all, Bond never fails. No one was going to tolerate a technology failure that might bruise his brand.Previously, HEINEKEN had supported digital media at its outsourced datacenter. But that datacenter lacked the computing resources HEINEKEN needed, and building them—especially to support peak traffic that would total millions of simultaneous hits—would have been both time-consuming and expensive. Nor would it have provided the geographic reach that HEINEKEN needed to minimize latency worldwide.
Case Study
Energy Management System at Sugar Industry
The company wanted to use the information from the system to claim under the renewable energy certificate scheme. The benefit to the company under the renewable energy certificates is Rs 75 million a year. To enable the above, an end-to-end solution for load monitoring, consumption monitoring, online data monitoring, automatic meter data acquisition which can be exported to SAP and other applications is required.
Case Study
Coca Cola Swaziland Conco Case Study
Coco Cola Swaziland, South Africa would like to find a solution that would enable the following results: - Reduce energy consumption by 20% in one year. - Formulate a series of strategic initiatives that would enlist the commitment of corporate management and create employee awareness while helping meet departmental targets and investing in tools that assist with energy management. - Formulate a series of tactical initiatives that would optimize energy usage on the shop floor. These would include charging forklifts and running cold rooms only during off-peak periods, running the dust extractors only during working hours and basing lights and air-conditioning on someone’s presence. - Increase visibility into the factory and other processes. - Enable limited, non-intrusive control functions for certain processes.
Case Study
Temperature Monitoring for Restaurant Food Storage
When it came to implementing a solution, Mr. Nesbitt had an idea of what functionality that he wanted. Although not mandated by Health Canada, Mr. Nesbitt wanted to ensure quality control issues met the highest possible standards as part of his commitment to top-of-class food services. This wish list included an easy-to use temperature-monitoring system that could provide a visible display of the temperatures of all of his refrigerators and freezers, including historical information so that he could review the performance of his equipment. It also had to provide alert notification (but email alerts and SMS text message alerts) to alert key staff in the event that a cooling system was exceeding pre-set warning limits.
Case Study
Coca-Cola Refreshments, U.S.
Coca-Cola Refreshments owns and manages Coca-Cola branded refrigerators in retail establishments. Legacy systems were used to locate equipment information by logging onto multiple servers which took up to 8 hours to update information on 30-40 units. The company had no overall visibility into equipment status or maintenance history.
