AIG Pensions
How I used primary research and user testing to ensure that a new site met the business goal of reducing calls to customer-service.
SNAPSHOT
Winter & Spring 2019: about 6 months
Business Partner: AIG Pensions
Product: A new self-service portal for AIG Pension holder (previously all service was via phone)
My Role: UX Research and Design
Outcome: The new portal, organized and designed based on research findings
THE OPPORTUNITY
As of late 2018, AIG pension owners wanting to learn about or manage their pensions had to call the customer service center, as there was no online access. With the goal of reducing back-office work, AIG’s Pensions Administration decided to build their first customer-facing site, and they turned to AIG’s Experience Design (XD) team to guide the design.
EVANGELIZING PROCESS
While the Business Sponsor and Product Owner were veterans in pensions administration, both were fairly new to digital product development, and neither were familiar with user-centered design nor Agile. The XD team knew that early and continuous education about both would be a key factor in the new team’s success.
A slide from the XD kick-off deck shows a high-level summary of our process for this engagement, which included HTML and CSS to best meet the needs of the off-shore vendor contracted for the development, as our growing pattern-library existed only in Sketch at this time (I co-authored the deck with Jordan Garegnani, XD Producer).
using research to test assumptions
IA as imagined by our business partners
Our business partners not only brought requirements, but also assumptions about our users’ online behavior (digitally naive) and the best way to organize the site (per current back-office system’s IA and work-flows, which center around “tasks” , requesting & submitting forms, SLA’s, and percent-completes).
To ensure the new site met user needs, I designed a survey and defined segments to learn how AIG pension holders’ online behavior compared to non-AIG customers of similar ages and pension-status, and the mental-models of both customers and non-customers, specifically around pensions-related content and management tasks.
The survey, built in UserZoom, included a card-sort exercise, a prioritization exercise, and a set of behavioral questions. Launched to both AIG and non-AIG pension holders, it received a total of 233 completes, with at least 25% of completes in each segment from users greater than 65 years old, the age-band of the majority of AIG’s active pension-holders.
I affinity-mapped findings from the user survey to identify key insights used to formulate the design strategy.
Patterns emerged from the results. We learned that:
All users, even older users, frequently manage finances online. In fact, 89% of both AIG and non-AIG active (retired) pensioners reported using a laptop or desktop for online banking within the past month. Design implications: users are familiar with common Web conventions and bring expectations to this new portal.
Users reported a preference for self-servicing online, but only when a website is easy to use. If content was difficult to find, or if the site required too many steps, users, and especially older users, preferred to call customer service. Design implications: to truly reduce calls to customer service, using the new site must FEEL easier than calling.
Users grouped self-service tasks contextually, with the other content related to the task (ie, Update Phone Number grouped with the rest of Profile-related data), not together with other tasks (ie, Update Phone Number grouped with Update Tax Withholdings and other tasks), as our business partners preferred to group tasks. Design implications: external users organize content and tasks differently than internal users.
I presented the findings and design implications to our stakeholders in an XD Vision deck. While there was some surprise, there was little push-back against this data-driven approach.
FINDINGS INFORM THE REQUIREMENTS
The card-sort data indicated that five total categories plus an alerts section would best help users navigate to the content and features they needed. These categories became our main navigation sections, and I worked closely with our Product Owner and our vendor’s BA and Scrum Master to organize the product requirements into JIRA epics for each main navigation section.
I created the site map above to illustrate the IA of the new site.
FROM FORMS TO FLOWS
As the research suggested, the new experience needed to not only allow users to self-service, but self-service options needed to feel as easy as possible. For certain tasks, like requesting a quote or an income verification, the easiest solution for our external users was likely going to be a bit of a departure from the input-field dense interfaces used daily by our customer service representatives.
I created wireframes (examples above) and a clickable InVision prototype to illustrate how we could use simple, dynamic screens to capture required data instead of mirroring the more complex and input-field heavy interface used by our customer-service representatives.
The wireframes I created served a few key purposes. One was to quickly demo directions and low-fidelity solutions to our business partners; another was to provide guidance to the Product Designer and Creative Technologist who were handing-off html and css to our off-shore vendor team (our design-system was immature and our pattern library not yet available in code); and lastly, as a guide to the API and data teams.
A/B TEST to DRIVE LANDING PAGE DIRECTION
In preparing for this project, the Pensions business team spent time documenting requirements and envisioning the new site, especially the landing page. Understandably, they believed a lot of information was worthy of such prominent placement, and were so used to the organization of their own systems that it was natural to want to carry-over those organizational themes to the new site.
Luckily, our early discovery and strategy work provided useful data and direction for the landing page design. Data from the call-center showed the most frequently requested updates, and data from our user-survey showed how users grouped information and the importance of the site feeling easy in order to meet the business goal of reduced calls to the call center. My hypothesis in designing the landing page was that the best solution would let users get directly to the task at hand, and show little data outside of that which users had ranked as very high priority in the survey.
Even though I had a data-driven hypothesis, it was still important to explore a few options, and I wanted to be sure to test an option based on our business partners’ experience with this user-group.
We created two versions of the landing page wires. On the left is a wireframe based on a very literal interpretation of the business requirements as written, while the one on the right was designed based on usability heuristics, data from the Pension’s call center about the most frequently-requested items, and information learned from the user-survey.
To determine the best version of the landing page, I conducted a static-click usability test. Users were asked to show where they would click to start four of the most frequently requested tasks, and then rate their experience based on ease and confidence level, and were assigned the above wireframes in random order.
A task asking users about where they would click to update their address was one of four tasks included in the usability test. Users were also asked to rate their confidence in each choice and ease-of-use after each task
Although both versions resulted in a fairly high success rates for the tasks, Version 2 resulted in higher confidence and ease-of-use scores for all tasks. I quickly reviewed the findings with the business stakeholders, the Product Owner, and our technology partners so we could move forward with the version that best met users’ needs.
THE FINAL DESIGN
Shaun Paduano, our Visual Designer, created comps with a combination of existing components from our growing pattern library plus some new components to meet the needs of the new Pensions site.
Unfortunately, AIG laid-off the entire Experience Design team a few weeks after we handed-off the final deliverables to our business and technology partners, so we were not able to collect any feedback regarding the new site once it went live or to participate in planning any next-steps regarding new / improved features.
It would have been interesting to track the effect of the new site on total calls to the call center and to explore ways to further help AIG Pensions reach their goal of increased online self-service.