AUTODESK
Using Content Strategy to Design a Better Customer Service Experience
Overview
DURATION
May 2018 - Aug. 2018
TEAM/ROLE
3 Content Strategists, 2 Designers, 2 Engineers / Digital Experience Intern
SKILLS
Content Strategy, Card Sorting, Website Taxonomy, Information Architecture
TOOLS
Drupal, Salesforce, Sketch
Presenting my project to my fellow interns and the VPs of Autodesk! (Click the image to enlarge.)
Context
Human customer service is expensive. With every call a customer service rep gets, their company is losing money and time. However, customer service chatbots are helping businesses save on customer service costs. Autodesk’s very own virtual assistant, AVA, has over 100,000 conversations with customers a month, relieving the burden of their customer service agents. But AVA has a long way to go: she can only recognize about 60 distinct use cases as of now.
This is where I came in as a digital experience intern at Autodesk: my job was to expand and improve the experience of AVA using the content on the Autodesk Knowledge Network (AKN), a rich repository of help articles and over a million contributions from the Autodesk community. My specific project within AKN focused on optimizing its help article pages to ultimately enable this content to be dynamically served up in automated experiences such as AVA.
The Problem
Goals
To understand and analyze key types of information that appear in Autodesk support content to ultimately improve the information architecture for their help website.
Create a tagging strategy that closes the gap between really long, frustrating information articles and solving the users’ problems/meeting their needs.
Design a successful pilot system that uses my tagging strategy to improve the UX of an automated customer service tool.
Defining the Problem
Users are frustrated by the lack of helpful, relevant responses to their questions to Autodesk’s Virtual Assistant (AVA), and aren’t willing to read extensive, unorganized support articles to find their answers.
Autodesk cannot currently use their support articles in any other context besides when users visit their website, because their articles aren’t structured or tagged in a way that can use modular content in automated support scenarios or experiences (such as the Autodesk Virtual Assistant).
The research phase consisted of analyzing over 250+ customer service articles on AKN, and interviewing potential stakeholders to start shaping my content strategy. I developed several insights from my research:
Research
Who are AKN’s users?
I recognized several types of users based on patterns I noticed in the support articles, as well as the types of information that these users come to the website for.
(Click the image to enlarge) Types of users and their use cases.
I started to see patterns in these use cases that helped me shape my revamped content strategy for better organization of the support articles. However, I also needed to understand the ways that my content strategy could potentially create more robust customer service experiences.
What are potential support scenarios for leveraging support content?
From talking to potential stakeholders within Autodesk, I discovered potential ways that support content could be repurposed:
→ Tooltips within Autodesk Account platform
→ Tooltips or other messages within Autodesk Store experience
→ Improving the UX of the Autodesk Virtual Assistant (AVA)
→ SEO
I ultimately decided to shape my content and tagging strategy to focus on improving the experience of the Autodesk Virtual Assistant, in addition to improving the way that Autodesk Knowledge Network’s current content is organized. In other words, my ultimate goal was to dynamically serve up impactful and relevant support content where and when the customer needs it.
Here is a map of the current user experience, which I hoped to improve:
(Click the image to enlarge) This is the current user journey for Project Manger, Jennifer. With a lack of organization in the content, the customer ends up frustrated and customer service ends up with more calls, which causes the company a loss.
Creating a Solution
(Click the image to enlarge) I started forming an information hierarchy with card sorting. I started with identifying different categories of information types. The goal is that all information in an article is organized by these information types, to give them a consistent structure.
(Click the image to enlarge) The next step was to develop content tags for different types of content that fall within these information categories. I took inspiration from the sorts of questions and phrases users may ask AVA regarding different types of information. Some examples include “getting started,” “how to…” and “issues.”
(Click the image to enlarge) I wanted to further develop the Guidelines information type, since this was the section that distinguishes use cases based on factors such as scenario and type of user. The final taxonomy tags would be added to the articles based on these types of details in order to provide accurate and relevant information to the right user.
(Click the image to enlarge) Finally, after developing over 20 different content tags within each information category, I tested the comprehensiveness of my content strategy by sorting modular content within each article to its appropriate content tag. The following picture is a preview of what this looked like, with just one section of support articles.
At the end of the testing process, I was able to sort the content in every support article to its appropriate content tag in my strategy, demonstrating that my taxonomy was complete and comprehensive.
Final Experience
Improved UX: Pilot Project Demonstration
On AKN, there are many troubleshooting articles containing solutions for different error messages a customer may receive when using Autodesk software, such as the one displayed below. However, the current experience isn’t ideal, as AVA’s response is to simply serve up a link to the whole article, which takes the user away from the experience:
Click the image to enlarge.
But with the help of tagging, we can tag the chunks of the article containing the “issue,” which is the error message, and the “solution,” or the step-by-step instructions on how to solve the error, and this is what happens instead:
When inputting the error message now, AVA responds with a more detailed response on the possible solution choices that the user can choose from. When the user clicks on one of the solutions, she takes you to exactly the part of the page containing the chosen solution, as she used the content type tags to identify the part of the page that the customer needs.
This improvement in customer experience demonstrates how powerful tagging can really be in unlocking the potential of automation, as this is the new user experience:
(Click the image to enlarge) With the improved performance of the Autodesk Virtual Assistant and my new strategy for organizing Autodesk’s help content, one step has been reduced in the user’s process of finding a solution.
Conclusion
During my time at Autodesk, I worked on a way that we can utilize content type tagging to make an impact in digital self-service as we know it. But what do we need to do next?
1. Explore other prospective support scenarios and experiences that could utilize content tagging.
2. Completely revamp and reorganize all existing support content on AKN’s articles with the content structure I have developed, and track their success metrics in terms of improving the customer experience.
3. Expand AVA’s capabilities further by tagging informational content with my taxonomy.
With the help of a more consistent content organization and a comprehensive metadata tagging strategy, we can enable support content to be used more effectively in automation, and ultimately work towards the future of digital customer service.