Tech & Innovation - February 21, 2023

Small steps and big bets: building an experimentation culture in Delivery Hero’s Product teams

Berlin, 21 February 2023 – At Delivery Hero, the world’s leading local delivery company, our product strategy is underpinned by the thesis that every day, our apps should be better than they were the day before. Experimentation is a key feature of this strategy: conducting large and small experiments on a mass scale enables constant progress. Small experiments focus on little steps, both back- and front-end, that improve the everyday functionality of our apps. Large experiments are targeted towards our “big bets” – tests that investigate new ways of building our products that will entirely change the way that customers engage with the platform. Both types of experiments are vital to the act of moving forward on a daily basis. So how do we build a culture of experimentation at a global tech company, and how do we ensure that every experiment is pushing us in the right direction?

Leveraging local learnings for global impact

Within Delivery Hero’s tech and product teams, we fail fast and learn quickly. Over the past year, we have been focused on implementing systems and processes that increase our speed of learning. The strategy is two-pronged: increase the number of experiments, and ensure the results are easily accessible. Delivery Hero’s unique strength is its global footprint. Operating across 70+ countries, with a large number of delivery apps running on multiple platforms, many experiments can be run simultaneously in different regions, with learnings then applied globally. By normalizing tooling and developing an experimentation archive, the Global Product Strategy team ensured that experiments aren’t being repeated on different apps or in different tech hubs. Instead, teams can quickly find past experiments and ensure that every test they conduct results in new data, from which new hypotheses can be drawn and implemented.

Creating a global culture

Conducting experiments is pointless if they are not consistent and accurate, which necessitates tools and processes that ensure these qualities. By prioritizing ease within our tools, we prevent experimental errors. As Delivery Hero, almost everything is built in-house, with the majority of experiments conducted via a tool called Fun With Flags. We also have groups within our teams that facilitate experimentation: the Experimentation Council focuses on the development of features that benefit all of the Delivery Hero Group’s platforms, while the Experimentation Guild looks at the outcome of experiments and offers feedback for improvement. Fundamentally, Delivery Hero believes in always aiming higher, which means that every team should be curious about the impact that they’re making on the experience of our customers, riders and partners. By baking this ambition into our global processes and structures, we continue to find new ways to delight our customers and push our platforms forward.

Rider tipping: a case study

A real-life example of the success of one of our experiments is focused on a feature that enables customers to tip their rider, with 100% of the amount going directly to the individuals. Different brands within the Delivery Hero group had a variety of different features, with varying levels of success, and some did not yet have the functionality. With the goal of incentivizing customers to tip their riders, Delivery Hero’s Product Strategy team connected with their local product teams around the world to map what each app was offering. They also conducted a competitor analysis. From this, the team designed a series of best practices for local brands to integrate into their apps, including offering pre-checkout tipping, a range of pre-set options for the tip total, and saving the customer’s previous tip, which enables them to repeat the same process with ease. In terms of results, one entity saw tipped orders increase to 20% after implementing the recommendations, and another saw tipped orders increase by 2000%. The results demonstrate the real-terms impact of experiments on our ecosystem, offering a simple route for customers to thank their riders directly.

Recent News