Blog

My team’s mission is to transform the Eventbrite support experience. I’m a senior content designer on the service design team: we help people who organize and attend events to navigate our products and troubleshoot problems. In the last few years, we’ve overhauled the design, information architecture, content, and user flows of the help content.  

Earlier this year, we introduced a new process that routes customers who contact support depending on the issue they need help with. My team sketched, tested, and iterated toward a simplified structure. As a result, up to 86% of customers can now accurately classify their support issues, which exceeded the team’s goals. So, how’d we do it?

The content design challenge

This project was part of the Eventbrite “Contact Us” experience: a complicated flow with conflicting stakeholder goals and user needs. We know that our users need to contact support quickly, but our stakeholders sought to add steps to the contact flow that would increase the users' mental load. We had to calibrate the friction to meet our business goals and satisfy our users. 

When event organizers who use Eventbrite contacted support, they weren’t required to provide much context before connecting to an agent. The problem with this flow was that all customer contacts got funneled into one big queue that all the support staff managed. When less-knowledgeable agents received requests that were too complex to handle, they had to transfer the customer to someone else. The longer the customer had to wait, the more frustrated they became.

The content design opportunity

Our goal was to enable Eventbrite customers to classify their support issues on the contact form to “route” requests.

The vision for routing: 

1. Help-seekers choose a topic from a menu that describes why they need help.
2. The help-seeker is routed to the right agent to solve their problem. For example, some agents are experts on payment issues, while others handle event registration questions.
3. The agent can solve the customer’s problem more efficiently because they have the knowledge and tools to take action.

Our success indicators: 

• Increase customer satisfaction scores
• Boost resolution rates (the agent solves the customer’s problem)
• Reduce time spent resolving each case

We created an information architecture to “route” customers who contact support. Routing was risky for the user experience because if the customer picked the wrong topic, they might not reach the right agent and have to wait for another agent to become available. Our customers might be hours away from their most significant event of the year—so every minute matters.

We took a long list of customer support issues and aimed to group them into an easy-to-navigate menu. Our hypothesis was the existing taxonomy of support issues was too lengthy and jargon-y for our users to navigate quickly.

Step one: Define the solution

The design team believed that if customers could accurately classify their support needs and the experience felt intuitive, then Eventbrite could provide higher-quality support and boost customer loyalty.

We redesigned Eventbrite’s existing information architecture of support issues, tested it, and launched an easy-to-navigate menu on the “Contact Us” form.

Information architecture is the way we arrange pieces of content to make sense when experienced as a whole. -Abby Covert

I started by aligning stakeholders and project contributors to agree on the following:

• The problem to solve
• The target audience 
• The metrics to measure success
• The timeline and deliverables

Step two: Research

We delved into existing user research to explore our customers’ needs, expectations, and attitudes.

• Web analytics showed many visits to the “Contact Support” page came from a help article. These visitors had already looked for self-service answers, but the fixes hadn’t solved their problem, and they wanted to talk to a support agent as soon as possible. 
• In-depth interviews revealed customers struggled to navigate our support menus. One customer said, “what I need help with is not on that list.” 
• Help article comments proved customers were panicked or crunched for time when they needed help. For example, if an event organizer’s funds are issued to the wrong bank account, they might not be able to pay their vendors on time. 

Given these insights, my team had to carefully calibrate the friction folks feel when they seek support. Asking additional questions about the issue may frustrate the user. Every experience must be clear and intuitive—easy enough for someone needing urgent help to breeze through. 

Step three: The design process

My role in the project was defining the problem, prototyping a new experience, testing and iterating the taxonomy, then executing and measuring it. 

According to HubSpot, “A taxonomy is the structure used for a website that organizes content in a logical manner so users can easily navigate the site and understand its purpose.”

I partnered with a few experts to create the first draft of our support issues taxonomy. We started by auditing the list of topics and subtopics our agents used to classify support cases—it was big, jargon-y, and did not reflect the audience’s mental models. There were dozens of topics, sorted seemingly arbitrarily. At a glance, I knew our customers wouldn’t understand topics like “payout reconciliation question” or “multi-event functionality.” 

So, we sorted the information into a new taxonomy. My partners and I aimed to clarify the topics and ensure our audiences could navigate them. 

First, the operations expert on the team helped me simplify the confusing subtopic labels to plain language. “Payout reconciliation question” became “where’s my payout?” We also consolidated similar subtopics. Then we compared our modified list to the navigation we use for Help Center articles–since we already did a tree test on the labels and groupings–so we were confident our audience would understand it.

Next, I led a design studio with a content strategist, customer support specialist, UX designer, and an operations expert. Our goal was to label and organize the list in a customer-friendly way. We experimented with editing the words and moving items to various groupings. Throughout two working sessions, we discussed all the topics and subtopics that didn’t align with the existing Help Center navigation. We asked questions like:

• How do our users talk about this topic?
• If this is different from the Help Center information architecture, should it be?
• Where does this feature show up in the product? How is it labeled there?
• How do our competitors describe and sort this concept? How do they approach “payments” versus “payouts"?

Lastly, I ordered our six topics and their 42 subtopics according to popularity. Data showed that event organizers requested help with “creating events” more often than “finance.” So, I placed it first in the menu. 

Step 4: User testing

Once we were confident about our first draft of the taxonomy, I conducted a tree test with our customers. According to Nielsen Norman Group, “a tree test evaluates a hierarchical category structure, or tree, by having users find the locations in the tree where specific resources or features can be found.” 

The testing goal was to observe how Eventbrite customers would navigate the topic and subtopic structure to classify their issues—and ensure urgent support issues were findable.

The UX designer on my team built a clickable prototype showing the new taxonomy. Then, we created a user test that challenged Eventbrite users to find solutions to 10 help issues within that taxonomy. For example, “Let's pretend you're using Eventbrite to organize events. You need help to add an administrator to your organization. Where would you go to learn how to add an administrator to your organization?” If they selected “Account” and “team management,” they completed the task successfully.

The recordings of real customers interacting with our designs showed stakeholders why testing was critical: some folks spent 3 minutes looking for a topic in the menu and couldn’t find what they were looking for. It proved that although we assume the things we design are simple and easy to use, our customers still struggle. I asked our stakeholders to imagine the customers trying to contact Eventbrite for help: how would they feel if they wasted 3 minutes navigating the routing menu on the email form?

Although some customers struggled, the tree test results showed most participants succeeded. Our customers found the right topics and subtopics for 70% of tasks. The team agreed the taxonomy was performing well, but we saw a few improvement opportunities. For example, our customers were looking for “promo codes” under “creating events,” not “marketing.” We studied where customers selected the wrong topic and iterated those groupings and labels. When we re-tested, the second version of the tree test achieved a 78% overall success rate.

Step 5: Launch and measure impact

Once the team was confident the taxonomy of support issues was clear, we implemented menus on the live chat and email support forms of the “Contact Support” page.

This project streamlined the tasks for the customer support team, enhancing their efficiency and empowering them to provide better support for our customers. Stakeholders appreciated that we brought their vision for routing to life.

Results after launch:

• Up to 86% of customers find the right subtopic to classify their issue and reach the right agent to resolve it. This outcome exceeds the team’s goal of 60% of users accurately classifying their problem.
• The new form fields did not harm the conversion rate: 86% of visitors who selected the email support form completed it. This rate is very similar to the baseline conversion rate.
• The project launched on time despite a few technical challenges.

Beyond the numbers, I’m most proud of how Eventbrite embraced collaboration and design thinking throughout this project. Our stakeholders trusted us to create and test a solution that would be clearest and easiest for our customers to navigate. They gave us the time and resources to design, test, and iterate—and they saw the value of content design firsthand.