AAA’s call centers were the primary lifeline for members in roadside emergencies, but the process was slow, expensive, and outdated. The existing digital alternative, a static "contact us" web form, left members in the dark. They submitted their roadside assistance request without knowing when, or if, they would receive a callback. The business mandate was to reduce call center volume and modernize the digital experience of requesting roadside assistance.

My goal was to transform this outdated roadside assistance process into a responsive, empathetic virtual assistant serving members in major markets like California and Texas.

By shifting from a static web form to a real-time chatbot experience, we provided immediate assistance to members and fundamentally changed how they access help. This work earned the AAA National Innovation Award for its impact on operational efficiency and member safety.

• Reduced roadside assistance request completion time by 85%.
• 60% increase in completion rate.
• $1M in annual cost savings.
• AAA National Innovation Award winner for transforming emergency roadside intake into a digital-first service at scale.

As the Senior UX Designer (lead designer on the project), I was responsible for the end-to-end conversational strategy and UI implementation.

I partnered with a dedicated UX Researcher to map the initial triage flows, focusing on simplicity for users in high-stress situations.

I worked with the development team to ensure our conversational flows were feasible within the constraints of the technical platform.

Collaborated with the Product Manager to define success metrics for the initial launch, establishing a process to monitor performance and prioritize improvements post-launch.

To understand the full scope of the service, I went beyond the office and conducted field research, shadowing Roadside Assistance Technicians during service calls.

Shadowing technicians during service calls revealed the friction on both sides of the experience. I saw the anxiety of members waiting on the side of the road, unsure if and when help was coming, and the frustration of technicians when dispatch provided vague location data.

Technicians relied heavily on visual landmarks rather than just GPS pins. This observation directly influenced our later decision to allow members to type their location (e.g., "by the Starbucks"), bridging the gap between digital coordinates and physical reality. This insight became the foundation for our hybrid location strategy, combining map-based confirmation with natural language input so the digital flow could mirror how technicians actually find stranded members.

Partnering with a UX Researcher, I helped create a comprehensive journey map to visualize the full service lifecycle. This artifact became our central source of truth, allowing the team to see the experience not just as a series of system processes, but as an emotional journey for the member. By overlaying user sentiment onto the technical workflow, we identified critical gaps in the current experience, specifically where the system failed to support members during their most vulnerable moments.

We architected the solution around three core pillars: designing for transparency, reducing cognitive load, and eliminating friction.

To counter the user's anxiety about "if and when help was coming," we prioritized immediate feedback loops. Unlike the legacy form, which felt like a "black box," the new flow provided clear, real-time confirmation that the request was received and dispatch was in progress.

I designed a hybrid interaction model that balanced speed with flexibility.

For standard inputs like "Flat Tire" or "Fuel", we used distinct quick action buttons to speed up the process of inputting information into the system.

For complex inputs like location, we allowed users to type freely (e.g., "I'm under the bridge near the exit"), leveraging NLP to extract supplemental location data that helped technicians clearly identify where the member was waiting.

Both the web form and early versions of the chatbot required a 16-digit membership number for member authentication. This was a major friction point, as few members had it memorized or carried their membership cards with them. I initiated and drove the push to replace this 16-digit membership requirement, collaborating with our Leadership team and the Membership product team to introduce a soft authentication flow using the member’s phone number, name, and ZIP Code. This single change removed the authentication barrier and significantly improved completion rates.

We designed the experience with a mobile-first approach, recognizing that the vast majority of users interact with the service via their phones during a breakdown. The responsive framework ensured seamless functionality across mobile web and desktop, and was later adapted and integrated into our iOS and Android native apps.

We moved away from rigid forms to a chat-based interface.

To reduce cognitive load in stressful situations, we prioritized clarity and minimal friction.

This included reducing barriers for entry (soft auth) and utilizing pre-set quick actions, which allowed users to navigate the flow with simple taps rather than complex data entry.

To execute the hybrid location strategy, I designed a flow that combined a visual map pin with a text field. Users could confirm their GPS location visually by dropping a pin on the map, or manually by typing "I am on 22nd Street, next to the Starbucks" for added clarity. This method directly addressed the difficulty technicians faced when relying solely on coordinates to locate a member.

Immediately following the launch of the new conversational flow, we observed a 60% increase in completion rate compared to the legacy web form.

We drastically improved efficiency, reducing the average time to complete a request from ~10 minutes (via phone call) to as little as 90 seconds (via the new chatbot).

By successfully diverting traffic from the call center and replacing legacy platforms, the project saved the company approximately $1 million per year in operating costs.

The solution proved highly effective and valuable during initial rollouts, leading to its immediate expansion into the company’s highest-volume markets (California and Texas) and integration into the native iOS and Android apps.

We recognized that roadside emergencies are high-anxiety situations where clarity and ease of use are essential. The conversational approach succeeded because it replaced a static form with a guided, step-by-step interaction that gave users real-time visibility into the process. This reduced the cognitive load for users and provided them with an alternative channel to request and receive assistance significantly faster.

Modernizing the static contact form into an interactive chat provided members with a crucial alternative to requesting roadside assistance. This new channel proved to be mutually beneficial: it drove higher engagement from users who preferred a digital-first experience, while simultaneously reducing annual operational costs by $1 million.

Feedback confirmed that the text-based interface naturally improved accessibility. By offering an alternative to phone calls, we made the service accessible to individuals with auditory impairments, eliminating the difficulty and stress of the traditional voice channel.

We regularly collected user feedback to monitor pain points and validate the experience. The response confirmed that the conversational approach was reducing stress and increasing accessibility:

"This is my first time I’m using the app, and it was very easy! It is stressful enough having a vehicle break down. This was very painless and not stressful."

"Great, way more accessible for those who are hard of hearing."

"Excellent, very easy user interface, far easier than calling!"

"First time using the chat, and it was an easy process...GREAT JOB."