An agentic AI service that empowers consumers in their customer service interactions
FairPlay AI provides consumers with the resources, knowledge, and support needed to resolve customer service issues in a quick, hassle-free, and effective manner.
Role
UX Researcher
Team
1 UX Researcher (me)
1 Product Manager
2 Designers
1 Business Consultant
Client
Consumer Reports
Timeline
8 Months
January - August 2024
Overview
The Why
Consumer Reports (CR) has been a trusted advocate for consumer rights since the 1930s, employing rigorous testing and its publication to promote safety and hold companies accountable.
Although CR continues to be a trusted resource for consumers during their purchasing process, its influence has diminished in an increasingly service-based economy. As consumer-company touchpoints extend beyond the initial purchase, CR's ability to safeguard consumers throughout the entire customer journey has become limited.
To address this, Consumer Reports approached us to design a solution that mediates interactions between consumers and companies to extend consumer protection and advocacy beyond the purchase stage.
My Role
I led user research and design strategy to create FairPlay, a mobile app and iMessage plugin that empowers consumers in their customer service interactions with companies.
I spearheaded user research, insight generation, and iterative prototype testing to inform the app’s core features and guide design decisions.
FairPlay AI is…
Solution
iMessage Keyboard Plugin
Mobile App
Voice AI Agent Service
Problem
Imagine you…
Impact
What Is?
Understanding the
Current State
Discovery // Understanding Client Business Needs & Objectives
Consumer Reports desires to expand its influence in the modern, service-oriented economy
Before diving into generative research to understand and define the problem space, we first sought to understand CR's strategic objectives. We conducted an Abstraction Laddering exercise with key stakeholders and uncovered the following underlying motivations.
-
CR aims to attract a younger audience and diversify its revenue streams beyond its current subscriber base, which has a mean age of 63.
-
Description text goes here
-
Description text goes here
What Is?
Understanding the
Current State
Discovery // Understanding Consumer Needs and Pain Points
Customer service interactions frequently leave consumers feeling frustrated and powerless. CR has the opportunity to step in to empower them during these challenging exchanges.
To expand CR's impact beyond the point of purchase, we first identified key touchpoints between consumers and companies, along with the needs and challenges that arise after a purchase. We conducted mixed-methods research—including netnography, contextual guerrilla intercept interviews, and directed storytelling sessions—to gather these insights.
-
We studied online consumer queries, discussions, and reviews on Reddit and company forums to identify the information and support that consumers frequently seek from companies after making a purchase, as well as the common pain points and challenges they encounter during these interactions.
-
We conducted 20+ contextual guerilla interviews at the Apple Genius Bar, Xfinity store, Target, and outside a Bank Of America to gather insights on the purpose of consumer visits and the common challenges they face when interacting with the companies post making a purchase.
This allowed us to narrow down on the most common reasons for interacting with businesses post-purchase, preferred channels of contacting companies, and consumer sentiments around online vs. in-person company interactions.
-
We conducted in-depth semi-structured interviews (n = 11) using directed storytelling and storyboard artifacts to capture specific narratives on consumer experiences in the post-purchase stage.
This set of interviews helped us explore in detail the journey, fears, motivations, and pain points that consumers encounter when interacting with companies.
My primary responsibility was to lead the design research process throughout the project. I selected appropriate research methods, designed research studies, developed study protocols, and crafted interview questions. I also actively participated in conducting the research with my team and led insight generation.
From our research, we uncovered that the most frequent and frustrating consumer engagements with companies occur during customer service interactions. These interactions are sought by consumers when they face issues with their product/service and require assistance from companies.
We found that customer service interactions are time-consuming, mentally draining, and emotionally exhausting for consumers. They experience feelings of helplessness and anxiety when interacting with companies to advocate for what they deserve.
Considering the challenges and insufficient support consumers encounter during this phase, we recognized it as a significant opportunity space for Consumer Reports to step in and empower consumers in their customer service interactions.
“Whatever it is, the way you tell your story online can make all the difference.”
What Is?
Understanding the
Current State
Synthesis // Understanding Consumer Needs and Pain Points
We conducted Customer Journey Mapping and Affinity Clustering to synthesize what we discovered from our generative research activities
“Whatever it is, the way you tell your story online can make all the difference.”
Synthesis // Insight Generation
We generated the following insights from our affinity clustering sessions —
Consumers spend a significant amount of time and effort in locating the right company contact information and gathering necessary documentation to support their case.
-
Consumers encounter difficulties finding the right contact information, gathering required documentation, and dealing with frequently uncooperative customer service representatives.
As a result, customer service interactions with companies are often time-consuming, effortful, and mentally and emotionally draining for consumers.
This presents an opportunity for Consumer Reports to streamline these interactions and make it easier for consumers to get the help they need.
“Whatever it is, the way you tell your story online can make all the difference.”
Consumers are limited in their ability to effectively negotiate with companies because they lack access to and understanding of company policies and consumer rights.
-
Consumers often do not seek out company policies and are unaware of their rights because of the overly technical, lengthy, and legal nature of policy documents.
This limits their ability to effectively negotiate with companies during customer service interactions.
Consumer Reports has the opportunity to present relevant policies in an easy-to-understand and actionable manner to aid consumers in building policy and fact-based arguments to effectively make their case during customer service interactions.
“Whatever it is, the way you tell your story online can make all the difference.”
Information and resource asymmetry causes consumer anxiety and low confidence during customer service interactions.
-
Consumers often operate as solo actors, lacking the time, expertise, and knowledge that companies have about their products, policies, and internal processes.
This resource asymmetry and lack of transparency leads to anxiety and diminished confidence when interacting with customer service representatives.
Furthermore, this asymmetry forces consumers to rely on emotional arguments and debates with representatives, making the interactions uncomfortable and emotionally draining.
This highlights the need for Consumer Reports to help increase consumer confidence in customer service interactions by mitigating resource and information asymmetry between consumers and companies.
Opportunity
These insights point to a solution that makes it easier for consumers to —
Connect with the right company representatives
Construct effective negotiation arguments using facts and policies
Reduce their apprehensions about interacting with businesses
What Could Be?
Imagining a Desired Future State
Ideation // Storyboards, Design Fiction Narratives, Paper Prototyping
To explore different ways in which the Consumer Reports can empower consumers in their customer service interactions with companies, we created storyboards, design fiction narratives, and paper prototypes.
Ideation // Parallel Prototyping
This allowed us to quickly explore different design solutions in various user contexts and led the way to three low-fi prototypes —
Negotiation Helper
Negotiation Helper provides —
A database of customer service numbers of different companies
Real-time assistance and policy alerts during customer service calls
Post-call summary, transcript, and tips
Policy Assistant
Policy Assistant is a browser extension and a mobile widget that —
Flags deceptive patterns and predatory behavior
Provides relevant policy details and insights in simple language
Drafts policy-backed arguments for consumers
CR Wallet
CR Wallet is a mobile app that —
Stores users’ products along with their respective customer handbooks, policies, and terms of service in a centralized “wallet
Help users troubleshoot product issues
Submits service requests on behalf of the consumer
What Could Be?
Imagining a Desired Future State
Ideation // Co-Design Session with Client
We decided to present our three initial design ideas to our client and conducted a co-design session with key stakeholders using the prototypes as tangible starting points to receive feedback, gain alignment, and get further clarity on client needs and expectations.
We conducted three activities, Rose, Bud, Thorn critique, Buy A Feature, and Build Your Own Product-Service during the co-design session.
The objective of these activities was to understand the motivations, concerns, considerations, and priorities of different stakeholders at Consumer Reports, from the Vice President of the CR Innovation Lab to the Technical Lead.
What Could Be?
Imagining a Desired Future State
Concept Testing // Round 1 - Prototype V1
TL;DR
Concept testing validated the need for negotiation support during customer service interactions, but users found real-time guidance during calls to be overwhelming.
While users valued the negotiation support, many preferred to avoid customer service interactions entirely; they just wanted their issues resolved.
When compiling client feedback, we found that a majority of key stakeholders prioritized Negotiation Helper’s real-time call assistance functionality and centralized company contact information feature, as well as Policy Assistant's ability to construct policy-based negotiations.
As a result, we designed a new mid-fi prototype where we combined functionalities from both design solutions.
This prototype integrates functionalities that address key consumer pain points and client priorities, identified through user research and client co-design session.
Prototype V1 had the following functionalities:
Compiled contact details of various companies in a phone book like interface
Real-time text support during the customer service call (our riskiest assumption)
Policy Insights
Call Summary and Transcript
-
In our first round of Concept Testing, I wanted to test:
Are real-time tips and policy insights effective in improving consumer confidence, customer service experience, and outcome?
Which functionalities are most/least effective in improving consumers’ customer service experience?
How is the simultaneous customer service call and app text support interaction for users?
-
I used Wizard of Oz testing in the first round of concept testing to simulate a customer service call and an AI agent within the app that provides real-time tips and policy insight notifications to the participant in response to the phone conversation. To start with, I formulated a scenario to put participants in context. Then, I tasked the participants to use the app to contact customer service regarding the scenario provided.
Scenario
Participant is overcharged on their Xfinity Wi-Fi bill due to a service change they didn't authorize.Roles
Moderator: Guides the participant through the test and observes their interaction with the prototype.
AI Agent (Wizard): Plays the role of the mobile app’s chat functionality behind the scenes, providing real-time information based on the conversation through text messages (performed by a team member in another room).
Xfinity Agent (Wizard): Plays the role of a customer service representative for Xfinity (performed by a team member in another room).
Participant: Interacts with the prototype and Xfinity customer service to resolve the overcharge.
01
Round 1 of Concept Testing validated the need for negotiation support during customer service interactions.
"Having the real-time tips is quite helpful. Because when the agent denied a full refund, the tips said I am entitled to a partial refund based on their policy. Without this guidance, I might have simply accepted the agent’s initial response." - Male, 20s
02
While participants valued the support provided during calls, most preferred to avoid direct interaction with customer service altogether. This highlights the need for a service that can resolve issues on their behalf without requiring their involvement.
“I would love it if CR would do the whole back and forth process - if I just entered my information, and they would do all the talking and figuring out.” - Female, 40s
03
Users found it challenging to process both negotiation tips from the AI agent and the ongoing conversation with the customer service agent. They expressed a preference for receiving policy insights and tips before the call, rather than during, to boost confidence and minimize distractions.
"I was confused about what was going on. I was trying to listen and then read at the same time and answer. It was overwhelming.” - Female, 50s
04
In-chat support was also identified as a valuable option for users who prefer chat-based interactions.
"I don't really like calling people on the phone. I feel like a lot of companies already have a chatbot and the chatbot can help you, so that kind of bypasses the need to call" - Female, 20s
I made the following design recommendations based on Round 1 findings —
01
Provide concise and direct talking points instead of long policy clauses and recommendations.
02
Present talking points before the customer service call rather than during it, to reduce overstimulation and help users feel more prepared and confident.
03
Offer real-time negotiation support for customer service chat interactions for users that prefer to use a chat modality to interact with customer service.
04
Automate customer service calls using a voice AI agent to address the problem behind the problem - resolving product issues without spending time and effort interacting with customer service processes.
What Could Be?
Imagining a Desired Future State
Concept Testing // Round 2 - Prototype V2
TL;DR
Preparatory talking points were utilized by users in their negotiation arguments, and they were effective in increasing user confidence and improving outcomes.
Automating customer service interactions with a voice AI agent was well-received by users. However, users found it cumbersome and stressful to takeover the conversation in the event of an AI error.
The design team implemented my recommendations to create Prototype V2, which introduces a new risky feature: a voice AI agent that calls customer service on behalf of the user. This functionality aims to address the problem behind the problem — users wish to directly resolve product issues without spending time and effort interacting with customer service. In Round 2 of concept testing, I sought to evaluate user reactions to this novel feature and assess its desirability and utility for consumers.
To address user feedback from Round 1, we also tested two interventions in Round 2: 1) pre-call talking points to reduce overwhelm and 2) a real-time in-chat negotiation support feature to cater to younger users' preference for chat-based interactions and empower them to advocate for themselves.
Prototype V2 had the following functionalities:
Voice AI agent (“CR Wizard”) calling customer service on behalf of the user
Take-Over feature allowing user to take control of the conversation from the voice AI agent
Policy-Based Negotiation Arguments
Real-time negotiation support during the customer service chat
Preparatory negotiation support provided before the customer service call
-
In the second round of testing, I wanted to test:
User perception of a voice AI agent
How do users perceive an AI agent acting on their behalf?
Does this feature add value to a user’s customer service experience?User response to AI errors
How do users react to errors made by the AI agent?
What impact do these errors have on user trust in the product?Preferred error correction modality
Which error correction method (text or voice take-over) is preferred by users?Preparatory pre-call talking points
How effective are pre-call talking points in preparing users for customer service calls?Real-time in-chat support
How effective are real-time talking points in assisting users during customer service chat interactions?
-
In Round 2, we employed Wizard of Oz testing to simulate both a customer service call and chat, as well as a voice AI agent making calls on behalf of users. The study involved two groups of participants who faced the same scenario: negotiating a replacement TV due to a picture quality issue.
Group 1 received policy-based talking points to prepare for a simulated customer service call. We aimed to test the utility of talking points provided before customer service calls. In part B of the study, participants were introduced to a voice AI agent (played by one of us) that interacted with a customer service agent on their behalf, initially without making any errors and then intentionally making a mistake to assess the participant's reaction and preferred correction behavior (text vs. voice).Group 2 also received policy insights, but in real-time during a live chat with a simulated customer service agent. We aimed to test the desirability of talking points provided during customer service chats. Similar to Group 1, participants were then introduced to a voice AI agent that initially interacted without making any mistakes. In the second phase, the AI agent again made a deliberate mistake.
To mitigate order bias, I randomly switched the sequence of Part A (where participants negotiated for themselves) and Part B (where a voice AI agent negotiated on their behalf) between different participants.
Overall, the study aimed to evaluate the effectiveness of pre-call talking points, utility of in-chat negotiation support, user reaction to AI agents acting on their behalf, and their ability to identify and correct AI-made errors.
01. Voice AI Agent
The majority of participants (80%) were comfortable with an AI voice agent handling customer service interactions on their behalf, particularly appreciating its ability to “advocate for them by referencing policies” and the “convenience of having an AI agent handle complex interactions.”
However, most participants had concerns about potential errors in the AI's responses; participants indicated they would only use the service if the AI demonstrated a high level of accuracy.
02. Perceived Value
Overall, all features tested in this round were considered valuable by users. When asked about their willingness to pay for the service, most participants (80%) indicated they would be open to pay for the voice AI agent service (an average of $26.40 per year). However, they expected policy insights and talking points to be offered for free.
Participants were generally hesitant to commit to a monthly or annual fee for the voice agent, as their need for the service depended on whether they faced customer service issues, leading many to prefer a pay-per-use model.
03. Talking points During Chat
From our initial tests, we found that younger users preferred text-based customer service interactions over phone calls. To address this, we introduced a feature that provided real-time insights from the AI agent during chat interactions.
This feature was generally well-received, although users found it cumbersome to constantly copy and paste the AI's insights into the chat box. To improve this, we redesigned the interaction to integrate insights directly into the chat box, eliminating the repetitive copy-and-paste process.
04. Preparatory Talking Points
Participants in Group 1 who received preparatory talking points with policy-based negotiation arguments before their customer service calls reported feeling confident and well-prepared.
However, since some struggled to remember specific arguments or policy clause numbers, displaying these talking points both before and during calls would be the most effective way to support customers.
05. Error Discovery
Only 50% of participants were able to discover the error made by the AI, as they often got distracted or multitasked while listening to the conversation.
This led to a desire for live transcripts that would allow them to monitor the dialogue and spot errors without a need to concentrate on the conversation.
06. Error Correction Interaction
60% preferred texting corrections over a voice takeover of the call, as they felt anxious about interrupting the conversation and feared potentially confusing the customer service representative.
While this method was favored, many found it challenging to type corrections quickly, leading to anxiety and confusion due to the lag. The remaining 40% opted for a voice takeover, finding it to be a more “natural” and seamless interaction.
I made the following design recommendations based on Round 2 findings —
01
For both text and voice interactions, I decided to add voice feedback from the AI agent every time the user attempts to communicate in order to facilitate a smooth transition and a confusion-free three-way conversation.
Text Interjection Example: “[Agent Name], the customer is typing additional information for this claim. Please hold for a moment.”
Voice Takeover Example: “[Agent Name], the customer would like to speak directly with you. Handing over the call in 3, 2, 1. Thank you.”
We tested these redesigned error correction interactions in the next round 3 of testing (below).
02
Direct Copy-Paste
03
Live Transcription
04
For Prototype V3, I decided to introduce pre-filled options for text interjection that users can select to pause the conversation between the AI agent and customer service agent temporarily so they can communicate corrections or additional information to the AI agent.
For voice takeovers, a designer on the group decided to redesign the interaction to involve a swipe gesture to minimize accidental interruptions, facilitating a more deliberate transition.
What Could Be?
Imagining a Desired Future State
Concept Testing // Round 3 - Prototype V3
TL;DR
Actively monitoring AI-led calls was time-consuming and demanded constant user attention, “defeating the purpose” of having an agent act on the user’s behalf. Interjection was awkward and stressful for users.
By providing pre-call verifications, in-call updates, and post-call summaries, users can passively monitor the AI agent, maintaining user control while maximizing efficiency and convenience.
Following the second round of prototype testing, I proposed design changes to facilitate a more seamless three-way conversation between the user, the AI voice agent, and the customer service agent. We specifically tested these design changes in Round 3 of testing using Prototype V3 shown below.
For this round of testing, I decided to embed a recorded conversation between the AI agent and the customer service representative within the prototype (see video below), rather than relying on live role-playing.
This approach streamlined the testing process, requiring only one team member per session instead of three, while also creating a more realistic simulation of an AI-customer service call because of the use of AI voice.
Prototype V3 had the following functionalities:
Voice AI agent calling customer service on behalf of the user
Live Transcript of the call conversation
Text Intervention feature allowing user to text the AI agent to provide additional information/correct errors etc.
Pre-written Options for text intervention
Voice Take-Over feature allowing user to take control of the conversation from the voice AI agent using a swipe gesture
Verbal Feedback by voice AI agent to avoid confusion (demonstrated in video below)
-
In this round of testing we focused specifically on the user experience of a voice AI agent resolving customer service issues on behalf of the user. Given the novelty of this feature and the potential for AI agents to make mistakes, we anticipated low user trust in the technology and a heightened need for user control. Consequently, we designed the feature to allow users to monitor AI calls and interject when necessary. This led to the development of the text interjection and voice takeover features tested in this round.
-
Present Problem Scenario (TV replacement)
Testing Error Repair Interactions
Participant is asked to resolve the TV issue using the Voice AI agent option.
Participant describes problem to agent via text/voice
Pre-recorded conversation between AI agent and customer service agent plays when participant presses “Act On My Behalf” option
AI Agent makes deliberate error during this conversation
Participant can either choose to interject via text or takeover the call
Text interjection
Takeover
We observed error repair interactions and ask follow-up questions on the experience
Research Findings —
While rounds 2 and 3 of testing validated the need for users to monitor the AI agent’s actions, round 3 revealed that they would like to do so passively, not actively.
All participants disliked the concept of a three-way conversation because it required them to actively listen to and monitor the AI-led call which was frustrating for users as it required constant focus and it “defeated the purpose” of having an agent act on their behalf in the first place. Many participants felt uncomfortable listening in on an AI conversing with a human because of the power imbalance in the interaction, and a majority felt anxious about needing to interject at any moment during the call.
Overall, while 70% of participants wanted the AI agent to engage with customer service agents autonomously, all users desired to receive confirmations along the way.
I made the following design recommendations based on Round 3 findings —
01
For both text and voice interactions, I decided to add voice feedback from the AI agent every time the user attempts to communicate in order to facilitate a smooth transition and a confusion-free three-way conversation.
Text Interjection Example: “[Agent Name], the customer is typing additional information for this claim. Please hold for a moment.”
Voice Takeover Example: “[Agent Name], the customer would like to speak directly with you. Handing over the call in 3, 2, 1. Thank you.”
We tested these redesigned error correction interactions in the next round 3 of testing (below).
02
Direct Copy-Paste
03
Live Transcription
04
For Prototype V3, I decided to introduce pre-filled options for text interjection that users can select to pause the conversation between the AI agent and customer service agent temporarily so they can communicate corrections or additional information to the AI agent.
For voice takeovers, a designer on the group decided to redesign the interaction to involve a swipe gesture to minimize accidental interruptions, facilitating a more deliberate transition.
What Could Be?
Imagining a Desired Future State
Concept Testing // Round 3 - Prototype V3
TL;DR
Actively monitoring AI-led calls was time-consuming and demanded constant user attention, “defeating the purpose” of having an agent act on the user’s behalf. Interjection was awkward and stressful for users.
By providing pre-call verifications, in-call updates, and post-call summaries, users can passively monitor the AI agent, maintaining user control while maximizing efficiency and convenience.
“Whatever it is, the way you tell your story online can make all the difference.”
Following the second round of prototype testing, I proposed design changes to facilitate a more seamless three-way conversation between the user, the AI voice agent, and the customer service agent. We specifically tested these design changes in Round 3 of testing using Prototype V3 shown below.
For this round of testing, I decided to embed a recorded conversation between the AI agent and the customer service representative within the prototype (see video below), rather than relying on live role-playing.
This approach streamlined the testing process, requiring only one team member per session instead of three, while also creating a more realistic simulation of an AI-customer service call because of the use of AI voice.
Prototype V3 had the following functionalities:
Voice AI agent calling customer service on behalf of the user
Live Transcript of the call conversation
Text Intervention feature allowing user to text the AI agent to provide additional information/correct errors etc.
Pre-written Options for text intervention
Voice Take-Over feature allowing user to take control of the conversation from the voice AI agent using a swipe gesture
Verbal Feedback by voice AI agent to avoid confusion (demonstrated in video below)
-
In our first round of Concept Testing, we wanted to test:
Are real-time tips and policy insights effective in improving consumer confidence, customer service experience, and outcome?
Which functionalities are most/least effective in improving consumers’ customer service experience?
How is the simultaneous customer service call and app text support interaction for users?
-
Present Problem Scenario (TV replacement)
Testing Error Repair Interactions
Participant is asked to resolve the TV issue using the Voice AI agent option.
Participant describes problem to agent via text/voice
Pre-recorded conversation between AI agent and customer service agent plays when participant presses “Act On My Behalf” option
AI Agent makes deliberate error during this conversation
Participant can either choose to interject via text or takeover the call
Text interjection
Takeover
We observed error repair interactions and ask follow-up questions on the experience
-
In this round of testing we focused specifically on the user experience of a voice AI agent resolving customer service issues on behalf of the user. Given the novelty of this feature and the potential for AI agents to make mistakes, we anticipated low user trust in the technology and a heightened need for user control. Consequently, we designed the feature to allow users to monitor AI calls and interject when necessary. This led to the development of the text interjection and voice takeover features tested in this round.
While rounds 2 and 3 of testing validated the need for users to monitor the AI agent’s actions, round 3 revealed that they would like to do so passively, not actively. All participants disliked the concept of a three-way conversation because it required them to actively listen to and monitor the AI-led call which was frustrating for users as it required constant focus and it “defeated the purpose” of having an agent act on their behalf in the first place. Many participants felt uncomfortable listening in on an AI conversing with a human because of the power imbalance in the interaction, and a majority felt anxious about needing to interject at any moment during the call. Overall, while 70% of participants wanted the AI agent to engage with customer service agents autonomously, all users desired to receive confirmations along the way.
These insights led me to recommend two design changes:
Provide Feedforward - Instead of requiring users to actively monitor calls and correct AI errors in real-time, users should be able to preview and edit the AI call script in advance and verify important details beforehand to minimize errors during the call.
Push Notifications for Passive Monitoring - Users should receive updates via notifications during the call, enabling passive monitoring of the conversation along with time-sensitive prompts from the AI at critical decision points. This was supported by an insight I extracted when testing the “text AI agent” feature — while users expressed discomfort with the idea of interjecting in a three-way conversation, many appreciated the ability to communicate directly with the AI agent. The push notifications and time-sensitive prompts facilitated this one-on-one interaction, allowing users to engage with the agent as it interacted with the customer service representative. Lastly, insights from earlier testing rounds highlighted the necessity of providing a call summary and transcript after each customer service interaction. This ensures users can verify the AI's conversation and maintain documentation of all their customer service engagements.
Favorability of Core Functionalities
I made the following design recommendations based on Round 3 findings —
Final Design Solution
The Solution
Reduces Time on Task
By recommending the most optimal communication channel and providing direct contact details, FairPlay reduces the time spent searching for company contact information and minimizes wait times.
By offering relevant policy insights for the issue at hand, FairPlay helps users avoid the time-consuming process of finding company policies and building a case.
With its voice AI agent handling customer service issues on behalf of the user, FairPlay eliminates the need for any active involvement in interactions with customer service, saving consumers time and effort.
The Solution
Streamline Issue Resolution
Consumers often find customer service processes tedious and lengthy, with many preferring to avoid these interactions altogether. They simply want their issues resolved without the hassle of dealing with customer service.
FairPlay streamlines the issue resolution process by automating it using different levels of AI intervention. At a lower level of AI intervention, it identifies optimal communication channels and provides tailored case arguments and talking points. At a higher level, FairPlay's voice AI agent can fully automate the entire customer service process, from initial contact to resolution. This streamlined approach saves consumers time, reduces stress, and ensures direct and efficient issue resolution.
The Solution
FairPlay AI
FairPlay AI is Consumer Reports’ agentic AI service designed to streamline and simplify customer service interactions for consumers. Acting as a personal advocate, FairPlay helps users resolve a variety of issues—such as billing errors, product replacements or repairs, subscription cancellations, and plan renewals.
It levels the playing field between companies and consumers by providing consumers with the resources, knowledge, and support they need to address customer service problems quickly and effectively. FairPlay’s primary goal is to reduce the time, mental effort, and emotional strain that consumers typically expend in lengthy and frustrating customer service exchanges.
The Solution
Facilitates Effective Negotiation
One common challenge consumers face when dealing with customer service is the need to build a strong case—either based on policies like warranties or through emotional appeals—in order to secure a favorable outcome. This process can be effortful and frustrating, especially when consumers lack access to or understanding of relevant policies and their rights.
FairPlay simplifies this by providing consumers with easy-to-understand, policy-based negotiation arguments sourced from company and legal databases. The intelligent agent extracts key policies and identifies potential loopholes to build effective cases for consumers. These are tailored to their specific issue and are presented as direct talking points to help consumers navigate customer service interactions with greater confidence and efficiency. Overall, by offering relevant negotiation arguments, FairPlay empowers consumers, reduces the effort and emotional frustration required to make their case, and levels the playing field with well-resourced companies.
What Could Be?
Imagining a Desired Future State
Ideation // Co-Design Session with Client
We decided to present our three initial design ideas to our client and conducted a co-design session with key stakeholders using the prototypes as tangible starting points to receive feedback, gain alignment, and get further clarity on client needs and expectations.
We conducted three activities, Rose, Bud, Thorn critique, Buy A Feature, and Build Your Own Product-Service during the co-design session.
The objective of these activities was to understand the individual motivations, concerns, considerations, and priorities of different stakeholders at Consumer Reports, from the Vice President of the CR Innovation Lab to the Technical Lead.
Synthesis // Co-Design Session with Client
From compiling stakeholder designs and prioritized features, we found that the following functionalities were considered the most valuable by a majority of key client stakeholders —
Centralized Company Contact Information
This feature resonated strongly with stakeholders as it addresses a common frustration among consumers of locating company contact details.
Policy-Based Negotiation Arguments
Stakeholders were enthusiastic about Negotiation Helper’s real-time text-based policy insights provided during customer service calls. This is because they believed it would provide great value to consumers while also holding companies accountable.
AI Agent Acting On Users’ Behalf
Stakeholders expressed mixed views on using agentic technology for automated customer service. While they recognized its potential to enhance value for consumers, they expressed concerns regarding costs and user acceptance.
Determining “What is”
Understanding the
Current State
Synthesis // Understanding Consumer Needs and Pain Points
We generated the following insights from our affinity clustering sessions
A significant amount of time and effort is spent in locating the right company contact information and gathering necessary documentation to support their case.
Consumers encounter difficulties finding the right contact information, gathering required documentation, and dealing with frequently uncooperative customer service representatives.
As a result, customer service interactions with companies are often time-consuming, effortful, and mentally and emotionally draining for consumers.
This presents an opportunity for Consumer Reports to streamline these interactions and make it easier for consumers to get the help they need.
Consumers are limited in their ability to effectively negotiate with companies because they lack of access to and understanding of company policies and consumer rights.
Consumers often do not seek out company policies and are unaware of their rights because of the overly technical, lengthy, and legal nature of policy documents.
This limits their ability to effectively negotiate with companies during customer service interactions.
Consumer Reports has the opportunity to present relevant policies in an easy-to-understand and actionable manner to aid consumers in building policy and fact-based arguments to effectively make their case during customer service interactions.
Information and resource asymmetry causes consumer anxiety and low confidence during customer service interactions.
Consumers often operate as solo actors, lacking the time, expertise, and knowledge that companies have about their products, policies, and internal processes.
This resource asymmetry and lack of transparency can lead to anxiety and diminished confidence when interacting with customer service representatives.
Furthermore, this asymmetry forces consumers to rely on emotional arguments and debates with representatives, making the interactions uncomfortable and emotionally draining.
This highlights the need for Consumer Reports to help increase consumer confidence in customer service interactions by mitigating resource and information asymmetry between consumers and companies.