Designing a new text search experience in GoFood
Designing a new text search experience in GoFood
Designing a new text search experience in GoFood
Every day more than a 1.2 million GoFood orders are placed using search. With close to half a million merchants on GoFood platform, it presents a significant challenge to quickly give results to the users, that are most relevant to them. As the senior product designer in GoFood, I was given the task to revamp the search experience of the largest food delivery service in Southeast Asia.
Every day more than a 1.2 million GoFood orders are placed using search. With close to half a million merchants on GoFood platform, it presents a significant challenge to quickly give results to the users, that are most relevant to them. As the senior product designer in GoFood, I was given the task to revamp the search experience of the largest food delivery service in Southeast Asia.
Process overview
Process overview
Understand
Understand
1. User behaviour - What did our data say?
1. User behaviour - What did our data say?
With millions of text search query done by our users everyday, internal data analysis can be very helpful. Quantitative data on user behaviour in the app always gives better insights on what users are actually doing over what users say they do in qualitative analysis. Hence, I collaborated with our business intelligence team to dig deeper into our internal data.
With millions of text search query done by our users everyday, internal data analysis can be very helpful. Quantitative data on user behaviour in the app always gives better insights on what users are actually doing over what users say they do in qualitative analysis. Hence, I collaborated with our business intelligence team to dig deeper into our internal data.
We took the top 100 queries among thousands and manually tagged them based on the user’s intent. Further analysis gave us four different types of query intents:
We took the top 100 queries among thousands and manually tagged them based on the user’s intent. Further analysis gave us four different types of query intents:
We took the top 100 queries among thousands and manually tagged them based on the user’s intent. Further analysis gave us four different types of query intents:
Dish intent:
Users search for a dish name
Brand intent:
Users search for a specific multi outlet brand name
Restaurant intent:
Users search for a specific restaurant name
Cuisine intent:
Users search for a category of food or a cuisine
Dish intent:
Users search for a dish name
Brand intent:
Users search for a specific multi outlet brand name
Restaurant intent:
Users search for a specific restaurant name
Cuisine intent:
Users search for a category of food or a cuisine
We looked at the repeat searches. On analysing the original to the final query in a booking session, we primarily got three reasons for re-typing or changing the original query:
We looked at the repeat searches. On analysing the original to the final query in a booking session, we primarily got three reasons for re-typing or changing the original query:
We looked at the repeat searches. On analysing the original to the final query in a booking session, we primarily got three reasons for re-typing or changing the original query:
Typo or different intent:
Final query is completely different from the original query or the original query had a typo
Identic:
Final search query is exactly same as the original query
Expanded:
Final query is an expansion of the original query
Typo or different intent:
Final query is completely different from the original query or the original query had a typo
Identic:
Final search query is exactly same as the original query
Expanded:
Final query is an expansion of the original query
2. Business needs - What was the business opportunity?
More than 65% all GoFood bookings were made from search. But, the search to booking conversion was very low. Search was very primitive in terms of features and capabilities at the time. Hence, improvements on search presented a significant opportunity to drive conversion and downstream effects of providing meaningful search experiences which can improve booking volumes.
More than 65% all GoFood bookings were made from search. But, the search to booking conversion was very low. Search was very primitive in terms of features and capabilities at the time. Hence, improvements on search presented a significant opportunity to drive conversion and downstream effects of providing meaningful search experiences which can improve booking volumes.
3. User Needs - What did the users say?
A series of user interviews were conducted with people who have used online food ordering apps to order or discover food. The candidates were from Indonesia and India. A careful mix of age, sex and order frequency was chosen. Key user insights from in depth user interviews:
Users start with a restaurant first & then look for dishes
Users search for dishes but expect a list of restaurants as results
Brands are associated with trust, quality and consistency of taste
Search is the primary mode of discovery on GoFood
Users know what they don’t want before they start searching
Users decide a cuisine before they start searching
Social media & recommendations by friends influence restaurant selection
A series of user interviews were conducted with people who have used online food ordering apps to order or discover food. The candidates were from Indonesia and India. A careful mix of age, sex and order frequency was chosen. Key user insights from in depth user interviews:
Users start with a restaurant first & then look for dishes
Users search for dishes but expect a list of restaurants as results
Brands are associated with trust, quality and consistency of taste
Search is the primary mode of discovery on GoFood
Users know what they don’t want before they start searching
Users decide a cuisine before they start searching
Social media & recommendations by friends influence restaurant selection
Define
Define
1. Key focus areas & goals
Based on the qualitative and quantitative data about user behaviour & needs we define the problem scope.
Help users make a decision at every step
Reduce search to selection time
Reduce number of repeat searches
Reduce cognitive load on users
Focus on restaurant funnelling
Reach search results faster
Based on the qualitative and quantitative data about user behaviour & needs we define the problem scope.
Help users make a decision at every step
Reduce search to selection time
Reduce number of repeat searches
Reduce cognitive load on users
Focus on restaurant funnelling
Reach search results faster
2. Hypothesis
2. Hypothesis
Ideate
Ideate
1. Design approach
Our understanding was that text search is not one step step but a journey in itself. The whole experience was broken down into 4 sub experiences. Each of these steps in the journey add to the overall search experience. It also helped to understand how search context moves from one step another.
Our understanding was that text search is not one step step but a journey in itself. The whole experience was broken down into 4 sub experiences. Each of these steps in the journey add to the overall search experience. It also helped to understand how search context moves from one step another.
Key Focus Area
Key Focus Area
Help users make a decision at every step
Help users make a decision at every step
Reduce search to selection time
Reduce search to selection time
Reduce number of repeat searches
Reduce number of repeat searches
Reduce cognitive load on users
Reduce cognitive load on users
Focus on restaurant funnelling
Focus on restaurant funnelling
Reach search results faster
Reach search results faster
Solutions
Solutions
Intent Classification
Intent Classification
Predictive suggestions, Query Understanding
Predictive suggestions, Query Understanding
Recent resto searches and not just queries
Recent resto searches and not just queries
Improved information hierarchy on merchant card
Improved information hierarchy on merchant card
Resto focussed dish results and new brand intent
Resto focussed dish results and new brand intent
Search context until merchant menu
Search context until merchant menu
Pre search: Predictive suggestions before user starts typing
During Search: Query understanding & intent classification
During Search: Query understanding & intent classification
No more empty states with spell check & auto correct
No more empty states with spell check & auto correct
New dish search experience within restaurant search & menu
New dish search experience within restaurant search & menu
Better information layout with redesigned merchant cards
Better information layout with redesigned merchant cards
