Delivery Hero and AWS: Cost Optimization in the Cloud

    Published on:

    delivery hero is one of the world’s largest regional delivery businesses, serving up to 2.2 billion people in over 70 countries. Its network covers 1.5 million restaurants and 1.4 million delivery drivers and couriers. It is the name behind well-known brands such as foodpanda, Glovo, foodora, and Talabat, many of which have been acquired over the last decade for growth.

    It also pioneered the concept of “quick commerce,” establishing the ability to deliver groceries and household items to homes within an hour (often within 10-20 minutes).

    Now, the company has become more focused on profitability and has turned to cloud computing to reduce the costs associated with operating such a large, global business. Similar to its rapid growth period when scalability was paramount, the company chose AWS as his cloud service provider primarily due to its global reach and the breadth of services it offers.

    This week I had the opportunity to speak with Michael Krenz, Director of Engineering at Delivery Hero, who works to manage cloud costs. We had the opportunity to discuss how global enterprises are leveraging cloud platforms to pursue profitability while focusing on innovation and customer experience.

    What is cost optimization and how does the cloud enable it?

    Growing a business on a global scale comes at a cost, as you continually expand into new markets and pour revenue into your business to attract new customers. But most companies come to a point where they have to prove they can make a profit. This period coincides with a period of economic uncertainty, as shareholders often seek reassurance that they are investing wisely.

    Cost optimization is one of the ways companies look to achieve profitability. This includes minimizing operational costs while preserving the customer experience.

    This is often cited as the driving force behind companies making the decision to move their operations to the cloud. This is because businesses can use the cloud to eliminate the need to purchase and maintain expensive on-premises servers and data centers, and streamline their IT infrastructure spending by moving to “pay as you go” and subscription models. This is because it can be Skilled and expensive staff were required to operate them.

    The situation is a little different for Delivery Hero, which has been cloud-native since its inception in 2011.

    Mr Krenz told me: “We have really relied on the wide range of services in his AWS ecosystem that have helped the company scale up and support the hyper-growth we have seen.

    “Both of our companies have a culture of innovation that makes them a great fit. AWS also has a very dedicated team of experts to help us find cost optimization opportunities. ”

    How did Delivery Hero achieve cost optimization?

    AWS cloud services have enabled Delivery Hero to undertake several initiatives aimed at reducing costs while maintaining service levels. One was to increase the use of “Spot Instances”, an AWS feature that allows customers to take advantage of spare cloud computing capacity (Amazon Elastic Cloud Compute) at a significantly reduced cost. Krentz said some teams have successfully migrated up to 90% of their workloads to these Spot instances, resulting in significant savings without impacting service delivery or customer experience. Also, he adopted a NoSQL database instead of the expensive and expensive SQL database that he had previously used for some operations, realizing further cost savings.

    An ongoing effort is a collaboration with AWS to create Deliver Hero’s own Customer Data Platform (CDP). This includes leveraging machine learning analytics across petabyte-scale customer datasets to better understand customer behavior and experiences.

    A separate system was developed to manage support and security checks across the company’s delivery drivers and passengers. It uses machine learning algorithms to match faces with ID card data to reduce fraud. Thanks to the responsiveness enabled by the AWS architecture, the service was deployed within his six weeks and reduced his fraud by 70%.

    Fraud detection was also key for another use case, this time the AWS Neptune graph database service. The aim here was to take advantage of first-time discounts for new customers by registering multiple accounts, thus reducing fraudulent customer losses. By implementing the system through Neptune, we were able to design a system that increased the volume of fraudulent purchases blocked by 32%. Importantly, by algorithmically identifying where multiple accounts are linked to one of her customers, she does this without needing to know who that customer is and without exposing personal data. It means that

    However, not all of the changes are purely technical.

    “Another big challenge in my opinion is the need to change the culture,” says Krenz.

    To do this, they needed to encourage their engineering team, which had traditionally focused on performance and growth, to view cost reduction as a primary business objective.

    he said to me “So when we find cost anomalies, we put a lot of effort into showing those costs right there, right away, not just on the dashboard, but through live notifications… To really motivate the team to work on it.”

    Overall, Krenz attributes about 10% operational savings to Delivery Hero’s cost optimization initiative, which is hosted on the AWS Cloud.

    “When you spend millions of dollars a month, even 1% savings can make a big difference,” he said.

    Next Steps in Delivery Hero’s Cloud Journey

    Delivery Hero will continue to innovate in the cloud with a mission to become the number one provider of local delivery services in all markets in which it operates. The company is also looking to emerging technologies such as AI to not only reduce costs but also create new customer experiences that allow it to differentiate itself from its competitors.

    The company already uses AI to predict customer behavior and optimize delivery routes, but it is also looking to a new generation of generative AI applications.

    Here, one potential use they’re looking at is automating the creation of menus and product descriptions across the 1 million+ restaurants they deliver from. If managed manually, it would be a labor intensive task requiring the creation of large amounts of images and text content. Generative AI has the potential to greatly speed up this process and make it even more efficient.

    Krentz told me: “We will continue to build more global services that local platforms can use so that we don’t have to reinvent the wheel.

    “And of course, we will continue to use every new service Amazon releases into the ecosystem.”

    To learn more about using Delivery Hero with AWS, please visit:

    you can click here Watch a podcast conversation with Michael Krenz, Director of Engineering at Delivery Hero. It delves deeper into different aspects of cost optimization in the cloud.


    Leave a Reply

    Please enter your comment!
    Please enter your name here