Seamless
food delivery
Seamless is an online food ordering service that connects users with local restaurants offering delivery or takeout through their website and mobile app.
I designed a new feature to increase engagement and decrease friction in their mobile app. This design challenge was conducted in the form of a five day design sprint.
My Role
User research
Personas + Storyboarding
Information Architecture
UX Copywriting
Wireframing + Prototyping
Usability testing
Challenges
As a Seamless user myself, I have long wished for the ability to leave private notes about restaurants that I could revisit later. Would my fantasy feature provide demonstrable value for other users? To find out, I needed to set aside my preconceptions.
I also had the constraints of working within an existing design and truncated timeline. All decisions needed to be coherent with the current product and of a scope that allowed for execution and testing within a five day period.
Process Overview
Task flow
Chat Prototyping
UX Copywriting
UI Prototyping
Usability testing
Affinity mapping
Iteration
Existing Product
I began by examining the Seamless app to develop a keen sense of the design patterns and functionalities in play.
-
Keyword search results can be filtered by delivery or pickup, or sorted by price, rating, distance, or speed of delivery
-
Sponsored and previously ordered results are presented first
-
Restaurants can be browsed by cuisine
-
Restaurant cards offer at-a-glance rundowns of key details, and lead to menu pages where users can add items to their carts
-
Primary navigation offers a perks page and account information
"I always order the same stuff. I should be a little less boring."
"I'm pregnant and my cravings are super specific."
"I order when I have people over and I don't want to cook."
"I'm generally operating under a time crunch."
"I look at Seamless when I don't have a recommendation from a real person."
User Interviews
I kept an open mind and guided conversations toward:
-
Discovering what motivates restaurant selection
-
Understanding points of friction or delight users experience when ordering meals
-
Determining where the existing product succeeds or fails in meeting user needs
-
Discovering environmental patterns users experience when ordering a meal
Empathy Map
To gain a deeper understanding of how Seamless users think and feel, I sorted their comments into an empathy map. Each color represents responses from one of the five potential users who participated.
User Persona
After analyzing data collected in the interviews, I synthesized my findings into a user persona.
Rob's motivations, struggles, and personality will guide my design decisions moving forward.
On day two, I identified the problem I would solve for Rob by adding a feature to the Seamless app.
While there was some evidence to support my suspicion that users would find value in making private notes about restaurants that they could look back on later, there was far more evidence that users wanted help being more adventurous in their ordering practices.
I concluded to set my preconceived feature aside and explored features that would empower Rob to try something new.
Defining the Problem
Ideation (Crazy 8's)
I set a timer for 8 minutes and sketched 8 possible solutions. The idea of a chat/conversational commerce function stood out.
Guided by a chatbot, Rob will feel less alone in his search and will be presented with a more manageable selection of curated restaurants than he would find using the existing search function.
Storyboard
Imagining Rob's environmental and emotional circumstances guided and inspired my design.
For example, Rob may be tired and hungry when chatting with Seamless. I spared him additional mental fatigue by providing a few quick-reply responses and the freedom to change his mind along the way.
Task Flow
I imagined a task flow for a typical interaction that Rob might have with the feature.
Rob's needs are a great indicator of the needs of a typical user. Prioritizing his task flow will provide a strong foundation for the product.
Prototyping
On day three, I prototyped the chat function. Guided by discoveries made while exploring the Seamless app, I created a prototype in Figma to show the interface design and how the feature would fit into the existing app.
To better demonstrate functionality, I also built a more interactive prototype in Botsociety. I gave the chatbot an upbeat personality to keep things light as it leads Rob through the path laid out in his task flow and storyboard.
Usability Testing
On day four, I presented my prototypes to a group of potential users and asked them to find a restaurant to order from.
I tested for discoverability and ease of use, noting areas of frustration or confusion, while ascertaining how successfully the design emboldened users to try something new.
I then grouped findings by affinity to discover trends.
“I love that there is a big sale sign right in front of me.”
“It was this cute little shop that I walked by all the time. Those places have the best stuff.”
“We needed it before our wedding and they didn't tell me how long it would take.”
“I don't know what demi-fine means, but it makes the jewelry sound cheap. Is that what it is?”
Findings + Conclusions
The general layout and navigation were a success. The feature was sufficiently discoverable and fit into the visual hierarchy of the existing product as intended. Many users expressed that the function would add value to their experience using the app.
There was room for improvement in generating feelings of trust in the function and building the confidence to type responses. Some users also wanted more information about restaurant pricing.
"I like the canned answers. You're not dealing with the anxiety of a blank page."
"I wish this existed! I would use this all the time."
"Chances are that what this showed me before would help me make up my mind."
"I don't have confidence yet that the bot can respond to what I type."
"I'd like to see more information about what any of these places cost. Like how much is this lo mein going for?"
Rebranding
I named the chatbot Simon and gave him a visual appearance that better reflects his friendly tone and functionality. Understanding the feature at first glance will build feelings of trust.
Fallback Scenarios
To address concerns with how well a chatbot can deal with typed responses, I considered fallback scenarios for unrecognized replies.
Since data from typed responses can help improve a conversational experience over time, I engaged the user with two opportunities to type something else before presenting them with quick reply buttons.
Suggestions +
Menu Pages
Users wanted to see more information about pricing, but Seamless does not display this information on restaurant cards in the current product.
I added a menu page to my prototype. Displaying the feature's suggestions first will help users quickly find the information they are looking for.
Updated Copy
Talking with users helped me to better understand how they converse and think when approaching the task of ordering a meal. I updated the copy to encourage more typed responses, and reminded users that they can scroll back up and select a different quick-reply path if they get stuck or change their minds.
**Botsociety was recently sunsetted.
Check back soon for a new prototype!
Looking Ahead
The next step would be to build additional paths and conduct more extensive user testing. After the feature has been launched, I would keep tabs on the product. Any patterns that emerge in how users interact and converse with the chatbot will provide valuable information as to how the experience could be improved moving forward.