Just Eat’s Smart Dish Search: Effortlessly Discover Your Favorite Dishes Faster

The Problem: Enhancing Choice without Complexity: Making It Easier for Customers to Find Their Favorites on Just Eat.

We understand that choice is crucial for our customers. However, as we expand the variety of restaurants and dishes on the Just Eat platform, it becomes increasingly challenging for consumers to find exactly what they’re looking for.

Customers who use dish search and get a successful result have a significantly higher conversion rate (44.48%) than those who do not get an initial successful result (23.38%).

We also know that 16% of those overall dish search users (404k) use the functionality, get an error, and do not get a successful result in the same session, these users convert extremely poorly (0.38%).

What Customers Were Saying

“Maybe this is a long way off, but the ability to search for a particular dish would be very helpful.”

“You have all the different types of cuisine, but I think you should be able to search for the food you want e.g. hotdog, steak etc.”

“Love the App. I use it fairly often, one thing I’d like to be able to do is search for a specific item e.g. Milkshakes and be able to see all the restaurants that sell them. Then I could compare menu variety and all that. Thanks!”

Leveraging Search For A Personalised Experience

While a search feature has always been part of Just Eat products and sees a good interaction rate, it has had some limitations. By enhancing this feature, we can gain a deeper understanding of our customers in real time. This enables us to create a more personalised experience by filtering out restaurants or cuisines they are not interested in and highlighting what they want. Combining search with customer ontologies allows us to be more accurate in what we display, positively impacting both adoption and conversion rates.

‘Delivering The Perfect Food Ordering Experience, Every Time.’

We saw this as a great opportunity to align with our overall company vision by addressing two key components:

• Be Human - Creating a simple, intuitive, and personalised experience that adapts to your needs.

• Be Habitual - Building a strong, unique connection that makes you feel like a valued partner.

Understanding Our Audience

We knew from our data that existing users were predominantly creatures of habit, same cuisine, same restaurant, same dish. Users new to our platform generally liked to explore options before making a decision. Just Eat already had several key archetypal customer types. We believed two especially would gain a lot of value from our ‘Search’ offering:

• Cuisines curious: Enjoy trying out new things and local independent restaurants

• Hungover & Hungry: Narrowing down on choice and cost

Our goal would be to provide a helping hand for these archetypes on discovering restaurants that addressed not only their cuisine taste but also their individual dish taste.

Going Beyond Basic Search

In analysing both direct and indirect markets, we observed a common trend: the reliance on basic ‘string’ search functions. Leveraging our deep understanding of the products sold on Just Eat’s platform, we saw an opportunity to go beyond simply matching search queries. By understanding the defining attributes of every product, such as spiciness, creaminess, or chicken content, we could introduce an intelligent exploratory element to search. This would provide a significant competitive advantage and elevate the customer experience.

Intelligent Matching Examples:

• Providing Reliable Options: Ensure that searching for “Creamy curry” returns dishes that meet this specific criterion.

• Smart Matching: Include results like “Tandoori Murgh Masala” when searching for “Tandoori chicken,” recognizing relevant dishes even if the exact term isn’t in the title.

• Alternative Suggestions: Offer similar dish alternatives if no exact matches are found, ensuring customers always have appealing options.

Workshop Activity: Finding Nirvana

I facilitated a collaborative workshop with stakeholders and team members to establish a common goal and identify potential obstacles. To define success, after a brief recap of market research, I asked participants to individually write a statement answering the question:

In 12 months, how would consumers talk about their dish search experience?

We then converged on a single statement that captured the group’s expectations for such an experience. This final statement became the foundation for our goal.

Vision Statement:

Just Eat understands my every craving. The experience is incredibly intuitive, allowing me to quickly find exactly what I want based on my unique tastes and effortlessly build a basket without ever needing to navigate a menu.

What could stop us from reaching our goal?

To ensure we achieve our goal, it’s crucial to recognise potential obstacles. The group individually brainstormed possible challenges and, through an affinity mapping exercise, we categorised these into the following themes:

• Not Discoverable

• Overly Complex

• Unrelated Search Results

• Lacks Key Decision-Making Factors (e.g., Price)

• Unclear Usability

• Menu Ontology Not Supporting Search Requests

Focusing On Personalisation And Key Challenges

The team recognised that personalisation was the cornerstone that would elevate our experience above the competition. However, we first needed to step back and gain a deeper understanding of two fundamental areas:

• Discoverability: How can we ensure the search feature is easily discoverable for every user who visits our platform?

• Complexity: How do we strike the right balance in the search results to avoid overwhelming consumers with too much information?

Answering Key Questions Through Discovery Research

To address these key questions, we conducted further discovery research, giving participants two tasks to better understand their expectations.

Defining Prototype Requirements

Based on our research findings, I collaborated with the Product Manager to develop a set of requirements to be included in a prototype for further usability testing:

• Intuitive Discoverability: Ensure the search feature is easily accessible at the SERP (search engine results page) stage.

• Predictive Search: Implement a predictive search function that suggests options as users type.

• Simplified Results: Display only essential information, such as price, ETA, and ratings/reviews, to avoid overwhelming users.

• Relevant Recommendations: Include helpful suggestions like “This dish goes well with…” instead of generic prompts.

• Alternative Options: Provide similar dishes or recommendations if no exact match is found for a search query.

Developing And Testing An Interactive Prototype

The next stage involved acting on insights from previous sprints to plan a more interactive prototype for user testing. This process helped establish the direction for a higher fidelity prototype exploration using Axure.

Usability Testing Insights

Overall Experience

Simplicity and Intuitiveness: Participants found the dish search prototype simple and intuitive, exceeding their expectations for a search feature.

Predictive Search

Positive Feedback: Participants were pleasantly surprised by the predictive results appearing as they typed, noting it was a great tool for avoiding spelling errors and recalling dish names.

Search To Menu Transition

Logical Placement: After selecting a restaurant and landing on the menu page, all users saw the searched dish at the top. They found this logical and appreciated not having to search through the entire menu, highlighting its intuitiveness and time-saving nature.

These usability testing insights confirmed that our prototype not only met but exceeded user expectations, providing valuable guidance for further development.

Iterative Design And Production

The design underwent several iterations of improvement before going into production. The first version of the dish search feature, which did not include ‘Predicted Search,’ achieved a 0.34% increase in good order conversions. After introducing the ‘Predicted Search’ feature, we observed a greater conversion gain of 0.42%. Considering Just Eat’s latest customer numbers for 2020 in the UK alone were approximately 60 million, this represents a significant revenue increase for the business.

Sharing Insights

Based on my experiences with this project, I wrote an article on setting a vision statement for the team at the start of a project. This helps avoid the common trap of stopping at an MVP, a challenge many teams face. You can read my post here on Medium