Jahez: Transforming
Missed Opportunities
Into Revenue Growth

Jahez, a leading food delivery platform in the Middle East with over 2.4 million daily users, recognized an opportunity to better utilize its search data to drive business growth. A substantial portion of its business — 55% of orders — comes from the app’s search function, which provides valuable insights into customer preferences. However, many of the restaurants users searched for were unavailable on the app, resulting in missed revenue opportunities.

To address this issue, Jahez needed a data-driven solution to identify these gaps and help its sales team quickly onboard the most sought-after restaurants.

Identifying Missed Search Opportunities

Before partnering with Dataiku, Jahez struggled to address missed search opportunities — instances where users searched for restaurants not yet available on the platform. Without automation, tracking and prioritizing these high-demand but unlisted restaurants was labor-intensive and slow, preventing the team from onboarding them efficiently.

Given that the search function accounts for more than half of all orders, capturing this demand was essential. With millions of weekly searches, manual processing was unsustainable, causing onboarding delays and preventing the sales team from keeping up with rapidly evolving food and beverage trends. To overcome these challenges, Jahez needed a real-time, data-driven system to empower its sales team to act quickly on customer demand.

How Dataiku Transformed Jahez’s Search Analysis

With Dataiku, Jahez revolutionized its approach to missed search opportunities. By automating the analysis of millions of search queries, the platform utilized advanced Natural Language Processing (NLP) techniques to identify unlisted restaurants. The transformation involved several key steps:

• Collecting Search Data: Each week, Jahez gathers detailed user search records, including clicked restaurants, user IDs, session IDs, cities, countries, and timestamps, providing a comprehensive view of customer preferences and search activity.
• Sessionization and Query Checks: Using NLP techniques, Dataiku enabled Jahez’s data science team to group similar search queries and match them to restaurant names. By applying sessionization, which turns event-based data into an ordered list of a user’s actions, the team tracked search behaviors more effectively.
• Filtering: Non-relevant searches, such as menu items, were removed to focus on restaurant names.
• Arabic-English Name Matching: To ensure both Arabic and English search queries were captured, an algorithm was applied in Dataiku to detect pronunciation similarities between the two languages.

The result was a solution that ranked missed search opportunities by the number of users searching for each restaurant, updated weekly. This provided Jahez’s sales team with a clear, data-driven view of which high-demand restaurants were missing from the platform, helping them prioritize which restaurants to onboard first. With millions of searches processed weekly, Jahez significantly improved both efficiency and responsiveness.

The data science team also benefited from automated pipelines in Dataiku. The Dataiku flow zones structured the project into clear, manageable steps, simplifying debugging and optimization. This automation ensured that as the volume of search data grew, the system could scale efficiently, making it easier to maintain and improve over time.

Dataiku brings multiple views of the data platform together in one tool, including orchestration through scenarios, building data pipelines, and AutoML. Having everything in one place has definitely reduced the distractions of switching between multiple tools, thanks to the seamless integration. Basel Alebdi Data Science Manager at Jahez

Unlocking New Levels of Efficiency and Revenue Growth

Jahez’s transformation yielded significant results across several key areas. By automating query management and data processing, Jahez empowered its team to process large volumes of data quickly and effectively, enhancing overall operational efficiency.

The business impact was significant, as the sales team successfully onboarded more high-demand restaurants, directly driving revenue growth. For instance, in Bahrain, one restaurant identified through the missed search opportunities system accounted for 45% increase in orders of newly added restaurants and 40% of total revenue generated from newly added restaurants within just four months. This shift from a manual selection process to a data-driven strategy greatly enhanced the team’s ability to identify high-value opportunities and improve operational efficiency.

Dataiku had a significant impact both technically and operationally. Technically, it improved project readability by breaking down pipelines into organized zones. With the power of Snowflake engines, we achieved almost instant results, with no network overhead. This integration greatly increased productivity, saving hours of development time. Operationally, SMO streamlined the manual process of finding restaurants to onboard. Now, the sales team has a clearer vision of which restaurants to target. Nouf Alroqi Data Scientist at Jahez

Why Jahez Chose Dataiku

Jahez selected Dataiku for its versatility in managing both data processes and project pipelines. The platform offered several key benefits that made it the ideal choice:

• Flexibility and Ease of Use: Dataiku’s intuitive interface allowed Jahez to create customized workflows and pipelines, boosting productivity and simplifying complex operations.
• Seamless Deployment and Version Control: Dataiku’s built-in features streamlined project management, enabling faster and more efficient deployment.
• Integration With Existing Systems: The platform integrated seamlessly with Jahez’s existing tech stack, including Snowflake’s Snowpark as well as MLflow, improving computational efficiency and enabling robust model monitoring.

Dataiku not only offers AutoML capabilities like some other solutions, but it also provides complete data pipelines, including steps for regular processing, ETL, and easy-to-use scenarios and deployments. That was a big plus. Finding a solution that combines data pipelines, AutoML, and collaborative enablement is a big plus compared to other tools. Basel Alebdi Data Science Manager at Jahez

The Power of Dataiku and Snowflake Together

The combination of Dataiku and Snowflake enabled Jahez to scale its operations seamlessly. Dataiku’s integrated workflows, coupled with Snowflake’s computational power, allowed Jahez to efficiently handle millions of searches weekly without performance bottlenecks.

By leveraging technologies such as Snowpark, Python, and SQL within the unified Dataiku platform, Jahez’s data science team built complex pipelines that significantly improved processing times and accuracy. This integration contributed to an impressive 92% reduction in query execution time, allowing Jahez to handle large datasets more efficiently. Additionally, Dataiku fostered greater collaboration through shared code snippets, reusable components, and project libraries, streamlining machine learning pipeline development and reducing redundancy across teams.

Driving Growth and Future Innovation With Dataiku

Through its efforts, Jahez has transformed restaurant onboarding into a highly efficient, data-driven system that captures missed opportunities in real time. By reducing query processing times and shifting to a data-informed approach, Jahez unlocked new growth opportunities.

Looking ahead, Jahez’s AI roadmap includes both micro and macro-level initiatives. On the micro level, the company is streamlining data processes and applying advanced techniques like NLP to address business challenges, such as delivery optimization and demand forecasting. On the macro level, Jahez is deploying advanced AI technologies, including the Dataiku LLM Mesh integrated with Snowflake Cortex models like Llama, Mistral, OpenAI, Cohere, and Claude. Currently, Jahez leverages Dataiku for structured data use cases, utilizes Snowflake’s LLMs for confidential data, and employs other models for general applications.

Generative AI will play a key role in advancing predictive modeling and personalization, helping Jahez better understand market trends and deliver tailored customer experiences. With these innovations, Jahez is positioned to lead AI-driven growth in the food delivery market.