Case Study: Designing an impactful experience
Improving conversion – architecting a chatbot conversation to ease friction in a two-sided marketplace.
The problem.
Reserve was a restaurant technology startup with a connected ecosystem of B2B and B2C technology products that facilitated a two-way communication network – allowing for real-time demand for a reservations to be disseminated from consumers directly to restaurants.
For restaurants – the request system allowed for a better understanding of customer demand for tables at given time and better understand, optimize, and manage their business.
For diners – the request system democratized the supply chain for tables at trendy restaurants, allowing them to ‘request’ a reservation directly and receive a yes/no answer, or potentially an ‘offer’ with alternate accommodations, sometimes within minutes, but often longer.
An unfamiliar process.
The request process did not align with diners established mental model around booking restaurant reservations. They were used to selecting from available times provided. Additionally, the static interface within the app that was being used to communicate information regarding the request process and the status of their “request” was confusing and frustrating for users.
Requesters were confused by the delay in receiving a response – Often resulting in them immediately cancelling their request.
Requesters were making assumptions that they had a confirmed reservation when the restaurant, in fact, had not approved their request – Resulting in some diners arriving at a restaurant without a reservation.
Requesters were confused & angry when confirmed for a reservation hours after beginning the process – Again often immediately cancelling their confirmed reservation.
The ask.
A previous phase of work that we executed that had included multiple rounds of iterative research & concept testing had signaled that a text-based, conversational experience would help to clarify the request process for our users.
The ask for this work was to bring an MVP version of the conversational experience to life for further testing & validation that would:
Set expectations – Clarify and provide transparency into the reservation request process.
Manage expectations – Set and manage a requesters expectations throughout the duration of the process.
Facilitate completion (conversion) – Allow Reserve to step-in to facilitate a positive outcome when a transaction was at-risk.
What would success look like?
Success for the redesigned experience would be measured by the following metrics.
Transaction completion (%) – An increase in the percentage of reservation requests that convert to a confirmed reservation.
Request cancellations (#) – A reduction in the number of reservation requests cancelled by the requester.
Offers accepted (#) – An increase in the number of offers (made by Reserve or a restaurant) that are accepted by requesters.
Reservation cancellations (#) – A reduction in the number of confirmed reservations cancelled within 12 hours of confirmation.
Delivery & outcome.
The final output of this project was a conversational experience and corresponding technology infrastructure that connected diners and restaurants, providing them with the ability to communicate within the system via a linked ecosystem of applications.
Understanding the physical world – The architecture and flow of the conversation was influenced by research into the interactions and conversations that were taking place between diners and restaurants in the physical world.
Making sure our experience was relatable – Insights from this research informed our design, allowing us to empathize with our diners needs and cater to these needs by structuring moments and language that was familiar and trustworthy.
Outcomes – By the numbers.
Through the use of a conversational experience that set and managed expectations, we were able to ease friction in Reserve’s reservation request process, resulting in positive movement in our key KPI’s over the first 3 months post-launch
While reservation requests remained within the app remained flat, the introduction of a conversational experience resulted in noticeable increases in successful request completion (%).
This was in part due to a decrease in unfulfilled request cancellations (#) and an increase in alternate accommodation offers accepted (%) that were directly correlated to the dialogue taking place via the conversational experience.
My role & contribution.
As the design lead on this project, I was responsible for leading, coordinating, and/or facilitating all aspects of design definition, design execution, and product management for this feature.
Design definition.
Collaborated with business and technology stakeholders to define goals, feature requirements & KPI's, coordinated customer discovery, and organized collaborative ideation activities.
Design execution.
Defined of system architecture, key flows interactions, iterative wireframe design UI & visual design, and the creation of specifications & documentation.
Program management.
Defined timeline, milestones & requirements, mapped user stories, set and maintained sprint cadence, facilitated discussions & decision making, and coordinated with engineering partners throughout development & QA.
Key components of the deliverable.
The key features of the redesigned experience were informed by a real-world interaction between a restaurant and a diner attempting to book a reservation. The experience facilitated a two-way conversation with diners (consumers) and a restaurant (business), with Reserve as an intermediary, helping to facilitate a desirable outcome (when necessary).
Submit a reservation request.
Let the restaurant know you are interested in a specific date and time so they can check availability.
Engage in a dialogue around your reservation request.
Send & receive messages from your Reserve concierge regarding your request.
Review alternate options sent by the restaurant.
Receive and accept or deny alternate time and seating options from the restaurant.
Receive alternate recommendations from Reserve.
Receive and accept or deny alternate restaurant options from your Reserve concierge.
Manage or take action on your request.
Manage your request on your own or interact with your Reserve concierge.
Get notified when your reservation is confirmed.
Know when your reservation is confirmed, view the details, and prepare for your dining experience
How we got there – 5 moments that mattered.
Getting to the final product was an iterative process that took place over a multi-sprint workflow – here are a few moments that mattered along the way.
Moment 01: Understanding
Determining what type of conversational interaction we should, and could build for our customers.
As a first step, it was critical to gather internal and external data that would help frame the problem, contextualize the landscape, and garner a thorough understanding of both the players and factors influencing the challenge ahead.
Desk research – Conversational landscape & technology audit.
Reviewing existing user research and analyzing comparative experiences provided us with an understanding of what was potentially viable and fostered our ability to have early, informed conversations with stakeholders.
Stakeholder interviews.
External – Discussions with front-line Reservationists at several of our key restaurant partners provided more clarity into what their typical conversations looked like when they were fielding reservation requests from diners in the physical world.
Internal – Discussions with Reserve’s technology & operations teams provided insights into existing pain points and what types of experiences would be feasible within the constraints of resource, technology and existing processes.
Synthesis & ideation.
Whiteboards, windows, and any surface in the war room were used to synthesize insights and capture ideas.
Organizing ideas into a system.
A mind map diagram was used to organize our ideas and visualize an initial, high-level approach to the system
Making initial decisions – What type of conversation should we build?
What should we build? Having this discussion with stakeholders and core team members early facilitated a consensus decision on MVP functionality.
Moment 02: Prototyping
Defining the conversation – iterative, low-fidelity prototyping.
Gator boards, post-its, sharpies, and tape were used to visualize and rapidly generate conversation flows for review & discussion.
Making decisions – What’s the scope of our MVP conversation?
Low fidelity flows and lightweight presentation artifacts were used to facilitate discussions, lock down an MVP scope, and align on a phased development plan.
Moment 03: Experience design
Defining the infrastructure of the conversation.
After determining the structure of our conversation, we focused our efforts on defining the details of the system – the interaction patterns and components that would bring the experience to life.
Message trigger types & scenarios.
Message triggers were the anchor components of the larger system. When triggered, they initiated a single message or a string of multiple messages
Conversation flow & ecosystem mapping.
A flow diagram was created to map all of the triggers and corresponding messaging flows within the system (including edge cases!)
Iterative interface & experience design.
Wireframes & ad-hoc user testing were used throughout the design process to zero-in on the optimal architecture for the interface & conversation modules within it.
Moment 04: Interface design
Defining the experience – bringing the conversation to life.
After defining the system, screen structure & content modules were designed, tested, then iterated upon to facilitate the necessary conversational user experience.
Custom content modules.
A core set of custom content modules were designed to present content, deliver prompts and facilitate interactions
Interface design.
Moment 05: Copywriting
Writing message copy – bringing the conversation to life.
As the messaging flow was being defined, I developed a simple process was to collect and track copy edits coming from stakeholders, across different teams. A spreadsheet was created using Google Sheets that mapped trigger events to all associated message strings and prompts.
This document allowed stakeholders to add content, make edits and leave comments without the need to decipher a complex user flow or requiring the design team to mock-up visual screens each time copy edits were made.
What I learned.
Through the use of a conversational interaction that mimics a real-world interaction between a restaurant and a diner attempting to book a reservation, we were able to design an experience that demystified a process that users were unfamiliar experiencing in the digital world – here are a few things I learned along the way.
Content-first is a powerful way to approach experience experience.
Discovery is still a requirement, but utilizing a collaborative, content-first approach to designing an experience provides a completely different perspective that generates unique ideas.
Conversations interactions helps us understand a users active intent.
Conversational interactions allows us to exchange information with a user. This exchange allows us to understand their context and active intent – creating an opportunity to deliver a personalized experience with the potential to surprise and delight.
User expectations rise sharply when interacting with a chat interface.
Chat interfaces are familiar to users and they are used to using them in a specific way – to communicate with in a two-way, open-ended, layered conversation. This mental model results in heightened expectations that are challenging to fulfill.
Human or machine? The line can be hard for a user to discern.
It’s important to consider how you want your chatbot to be perceived by a user – as a machine or another human. But even then, it may be hard for users to discern what they are interacting with, don’t be surprised by how many people will respond with “thank you!”. That’s a good thing though.
Conversational interaction patterns can be utilized in many ways.
Conversational interaction patterns can be leveraged in a multitude of ways. By thinking differently about how they can be used, we can design creative solutions that push conversational interactions in novel directions.
Want to know more?
Have questions? Want to learn more about my experience architecting a chatbot conversation to demystify an unfamiliar process and ease friction for Reserve users? Contact me.