Four interlocking puzzle pieces

Optimizing Bank of America’s mortgage experience: From a fragmented journey to record-breaking user satisfaction

My role: UX strategist, information architect, interaction designer, and lead for a 4-person design, research, and content team.

Screenshot of the primary site

Primary site: Home Loans Sales

Screenshot of the standalone site

Standalone site: Home Loan Guide

Directing traffic to a dead-end street

Bank of America Home Lending was missing their digital sales targets. We discovered the bank’s marketing agency had created a first-time buyers Home Loan Guide, and was directing all paid search and campaign traffic to this standalone site, instead of the “official” mortgage buying experience. The standalone site had no other content and no CTAs, causing drops in sales and user satisfaction.

As UX lead for consumer lending, I facilitated conversations across siloes to align on common ground: We needed to meet user needs throughout the mortgage journey, improve conversions, and reduce the cost of operating multiple sites.

Working in a challenging context

The project landscape was complex, requiring careful navigation among 50+ individual contributors, 23 stakeholders, 14 executive approvers, and 6 siloed departments. In addition to critical requirements from marketing, sales, legal, and compliance, engineering was launching a new CMS, brand launched a new style guide, and we (UX) launched a new design system for bankofamerica.com.

We also had skeptical business stakeholders who had never done user research. They firmly believed experience decisions should be made by product owners and multivariate testing. They were certain that talking to users would skew results to meet a hidden agenda. To reassure them, we included them as contributors, approvers, and observers in all research activities. We also recruited a high number of research participants. This helped our stakeholders feel confident that user research insights were valid.

Establishing guiding principles was critical

With competing priorities and conflicts among our siloed stakeholders, I quickly realized we’d need clear guiding principles to help keep work on track:

  1. Prioritize stakeholder engagement

  2. Lead and validate thoroughly with extensive user research and testing

  3. Put user education first to ensure a safe, trustworthy experience

  4. Make products and pricing clear and visible to support SEO and sales

  5. Leverage new brand and design systems

  6. Allow for future phase concept exploration

Finding out what users really need

I interviewed key stakeholders and subject matter experts, reviewed user analytics from current sites, examined previous user research insights, and learned the end-to-end mortgage sales processes.

Simultaneously, we conducted extensive usability walkthroughs and user interviews with 40 participants (covering multiple personas and scenarios), with a goal of validating assumptions, and discovering unknown needs, preferences, and behavioral drivers.

From this foundational research, and underpinned by our existing consumer banking personas, five behavioral archetypes emerged, with Novice and Expert behaviors (the two most common) as our highest priority:

Illustration of the expert archetype
Illustration of the novice archetype
Illustration of the customer archetype
Illustration of the in process archetype
Illustration of the distressed archetype
Description of the expert archetype
Description of the novice archetype

Streamlining the information architecture

I comprehensively inventoried the content and features from both sites, and then organized them into cohesive categories, solving for redundancies. With consideration to our guiding principles and priority archetypes, I created a preliminary integrated information architecture.

Table listing the sources and types of information in the IA

Receiving positive user validation

Next, we conducted open card sorting and sample content evaluations to inform and validate this IA, as well as the potential to repurpose existing content.

All samples of existing content garnered positive responses and could be repurposed with minimal editing. This was particularly good news, as it expedited legal and regulatory review.

I had a clear understanding of business and user goals. The scope and approach to content integration had been confirmed by real users. I could confidently move forward to designing navigation.

Collaborating on navigation

In combining the sites, we repurposed the concept of process steps as mandatory navigation, in order to preserve the standalone site’s approach to Novice education. 

Diagram of process based navigation requiring users to select a process step and their need for guidance

We needed effective secondary navigation design for process steps. But, the new design system did not have secondary navigation patterns. Through joint design sessions with the marketing agency, we co-created three secondary navigation concepts: Wizard, Accordion, and Filmstrip

Wireframe of the wizard navigation concept
Wireframe of the accordion concept
Wireframe of the filmstrip concept

Seeking user input on a key navigation decision

This navigation design was a high-risk decision. We decided to prototype and test the options with users. We quickly pulled together 3 mid-fidelity prototypes, each with about 40 screens, and structured the test to evaluate:

  1. Relative learnability and usability of the competing concepts

  2. Ability to choose between guided vs. self-directed experiences within each step in the process.

  3. Other qualitative user feedback on layout, content, ease of use, and overall satisfaction.

Wireframe of the design system template
Wireframe of the wizard prototype
Wireframe of the accordion prototype
Wireframe of the filmstrip prototype

Uncovering a big miss

Our testers found that all three options provided a generally usable site that prioritized their needs. They were satisfied with the visibility of rate information and tools. Plus, they felt product information was easy to access, with an appropriate level of detail.

But, having to choose between Novice and Expert experiences was a significant usability failure. When presented with that choice, users could not accurately predict if they would need guidance. For those that said they were certain, their actual behavior proved otherwise. 

Removing high-friction decision points

With this crucial insight, I simplified the structure, eliminating the need for users to choose process steps and knowledge paths, and instead enabled flexible access to all potentially relevant content. Key information and tools were also made more visible by moving them up in the hierarchy.

From this:

Diagram of the unusable site structure

To this — users no longer have to choose a process step or predict their need for guidance:

Diagram of the new site map

Creating detailed design specs

Now that the site structure and navigation were locked, I dug into the details. Because I was using existing content, existing functionality, and the new design system, I decided to produce standard layout templates for each page type, referencing the repeatable content and components. This approach would speed up high-fidelity design and development.

Evolving CTA design and function

I needed to create net new components for dynamic CTAs, menu-based CTAs, marketing campaign integration, and contextual navigation.

The CTA component was complex, displaying any permutation of 5 different channels: phone, chat, search for a local salesperson, apply online, prequalify online. Business rules were applied to dynamically change the component at the page level, so that traffic would go to the most optimal channel(s), in real time

Adding prominent CTAs

Business stakeholders were concerned the integrated site had too few CTAs. Because the primary navigation was now so simple and straightforward, I took the opportunity to add an extremely visible and persistently findable CTA.

Exploring contextual navigation

As users reviewed articles and videos, I wanted to offer easy ways to identify content they had already viewed, content we recommended they view next, and the content format (text or video).

I designed contextual navigation to keep users engaged, with a clear path to related content. 

Annotated mockup of static contextual navigation concept
Annotated mockup of future state dynamic contextual navigation concept

Concepting multi-session experiences

User research had also revealed that people expected this process to be complex and time-consuming, and anticipated the need to make multiple visits over time. I provided inspirational designs for a future My Home Loan Folder feature that could track and communicate progress. User must-haves for the experience included: retrievable over multiple sessions, helpful in completing the application, visible status and next steps, and shareable with partners in the process.

Mockups of the future state folder concept

Getting positive results from final usability testing 

We conducted one last round of usability testing, with an objective of determining if the structure and navigation was intuitive and usable, and the ability to complete 12 specific tasks for each of the product categories. The results were definitively positive, with only minor issues that could be safely revisited after launch. 

Preparing final specifications

Calculators and other tools needed to be re-engineered as part of CMS migration, but there was no existing functional spec. I worked closely with subject matter experts and engineering teams to completely document all the business rules, so these features could be coded and tested.

During this project I developed a close and trusting relationship with the engineering team, particularly the QA lead. Our daily conversations gave me insight into her biggest pain point: no single source of truth. This was the pre-Jira era, so I devised and maintained a spreadsheet that included, for every screen:

  • Links to the wireframe, component specs, assets, copy deck, and functional specs

  • Metadata

  • URL redirect mapping from the two existing sites

  • UX, dev, and QA status

Running the approval gauntlet

We needed approvals on the final designs from 14 different executives across 6 different silos, each with their own process and criteria. As UX team lead, I navigated and negotiated to get these boxes checked. 

The only notable glitch was the accessibility review. Due to technical constraints, screen readers could not interact with overlays nested inside other overlays. Thankfully, this was a rare occurrence (a few glossary terms inside calculators), and it was simple enough to remove the extra overlay.

The result: Satisfied users drive better business outcomes

  • 5x increase in CTA pull-through

  • Annual online mortgage sales exceeded $1 billion 

  • Record high customer satisfaction scores 

  • Ranked #1 in digital sales functionality (Forrester)

Mockup of the integrated home page

Looking back on a growth opportunity

Integrating the home loans experiences was a long and challenging journey. I was able to practice and deepen important skills:

  • Staying focused on our users

  • Prioritizing information architecture

  • Executing with speed and quality

  • Guiding and supporting a multi-disciplinary UX team 

  • Managing stakeholders across a large organization

  • Developing and applying business knowledge

  • Collaborating with engineering

  • Having grit and perseverance through it all

Putting people first

This was a large and complex project inside a large and complex organization. In situations like this, I’ve found success in:

  • Providing a clear and consistent vision that everyone can understand and connect with.

  • Treating stakeholders like users, focusing on their needs, pain points, and delights, and trying to make their experience of working with UX as delightful as possible.