At this year’s Delivery Hero Hackathon, the team “The Review Revolutions” secured first place with their fresh, AI-powered approach to customer reviews and ratings, which began with the question “How can we improve user-centricity and navigability of browsing customer reviews and help you decide quickly what to order?”
The theme of the 2023 Delivery Hero Global Hackathon was “Igniting growth with AI”. Participants from around the world formed teams to ideate, execute and present their ideas within a span of just three days.
Our team—The Review Revolutions—was incredibly diverse, bringing together Heroes from three different entities – foodpanda, Delivery Hero and Pedidosya. Despite being spread across various locations across the globe, our shared objective united us: to leverage the power of AI, particularly Large Language Models (LLM), to provide consumers with rapid, in-depth insights from thousands of reviews. It’s like giving them the essence of every review, making their decision-making faster and more informed.
The Review Revolutions: Empowering Users to Make Informed Decisions Through Review Summaries and Sentiments
Revolutionizing Ideas: The Journey Begins
As we dove into making restaurant reviews more user-friendly, we kicked things off with the simplest idea: giving users quick summaries of all related reviews. But that was just the beginning.
We immediately explored other ideas, including linking reviews to ordered items, extracting quantifiable data points from the reviews, and brainstorming creative ways to summarize them. We took inspiration from industry leaders like Google Maps and Amazon to see what they were doing right.
Our final goal? To guess potential customers’ questions and have the perfect answers ready, supported by relevant customer reviews.
Executing on the Vision: Leveraging LLMs and the Team’s Strengths
For the implementation, we experimented with various Large Language Models (LLMs) capable of handling many reviews in multiple languages. Finally, we chose to use Chat-GPT from OpenAI as it allowed for rapid iterations without additional training of the models.
All customer reviews were submitted to our AI pipelines, grouped by vendors. Each review was meticulously analyzed to extract favourable, yet honest summaries, common topics about vendors’ characteristics and various sentiment analyses. By identifying common topics as tags, each review was classified based on its relevance and sentiment related to those tags. After completing this process, the results were compiled into a comprehensive review analysis and stored in our backend. The front end then displayed these results in an interactive and insightful manner to the users.
We aimed for quick feedback cycles while ensuring end-to-end integration from real customer reviews right through to front-end mocks. Our team’s diverse composition, spread across Berlin, Singapore, and Argentina, helped bridge significant time differences.
For instance, our designer in Singapore, with a 6-hour head start, was able to create innovative product design ideas. These concepts were then passed to our members in Berlin, consisting of a Backend and Full-Stack Engineer, a Data Analyst, and a Data Scientist, who transformed these ideas into tangible implementations. Following this, another 5 hours later, our Frontend Engineer in Argentina joined in to integrate these developments into the frontend. This structure enabled us to maintain a daily innovation cycle, perfectly timed to meet our prototyping needs.
The Magical Outcome: The Customer Experience
Despite our ambitious goals and the tight timeline, we successfully delivered on our vision with a fully functional end-to-end prototype. This prototype demonstrated the true potential of the ideas. Whereas previously, customers could only see a star rating and an endless list of individual reviews, they were now presented with a concise summary and the option to delve into various vendor-specific aspects. This allowed customers to make quicker and better decisions by immediately understanding the sentiment on the aspects they value most.
The impact of our solution extended beyond just the customers. It also gave vendors a deeper understanding of their customers’ experiences, going beyond the individual stories in the review.
The Cherry on Top
The entire endeavour was not only enjoyable and engaging but also an enlightening experience for all of us on the team. The icing on the cake was the recognition of our dedication, collaboration, and willingness to embrace challenging ideas during the hackathon. Our project caught the attention of a distinguished panel of judges, leading to our selection as the hackathon winner.
If you like what you’ve read and you’re someone who wants to work on open, interesting projects in a caring environment, check out our full list of open roles here – from Backend to Frontend and everything in between. We’d love to have you on board for an amazing journey ahead.
