Actionable Insights for WiFi Management

Comcast Business offers a “one-stop shop” business internet page for small-to-medium businesses to easily view and manage their various internet services.

Plan Upgrade Recommendations was a net-new feature made to help businesses see if their current internet plan met their business needs, and push to upgrade if it didn’t.
UX DESIGN, CONCEPT TESTING - 2022

The Challenge

How might we ensure that small to medium businesses will always have an internet plan that supports the bandwidth needs of their internet users?

The Stakeholders

I was brought onto this feature knowing the ask and what my stakeholders wanted. But what did customers have to say?

Validation Through Concept Testing

I determined that concept testing would be a quick, low-cost way to get real user feedback on whether customers wanted this feature before we sunk in the time to make it happen.

Deciding the Feature Location

Before I started iterating on the design concepts to test with, I had to decide where the recommendation should live. I didn’t want the testing results to be influenced by the positioning of the content in relation to the page. After a pros and cons analysis, I decided to place Plan Upgrade Recommendations in the "My Plan" section (Option 1).

Current State Business Internet page with two placement options for this feature, Option 1: My Plan or Option 2: Data Usage.

Ideating the Concepts

Then, I designed three versions of this feature to see what would best resonate with users. These variations targeted the following research questions:

How useful and insightful would customers find this plan upgrade recommendation to be?
How likely would a customer attempt to upgrade their plan based on this recommendation?
What kind of content is needed to support this recommendation for a user to feel that this recommendation is trustworthy?

Version A: Simple

Version B: Simple, with details on click

Version C: Data and details showcased upfront

Getting the Results

I recruited 5 participants through UserTesting to find existing Comcast Business customers that had familiarity with this Business Internet page. Each user was asked to speak out loud and interact with a clickable prototype that showcased the concepts one by one before viewing a comparison of the three concepts side by side.

Comparison of the original card design against the 3 concept versions.

After the test was complete, we compiled the results and presented a readout to our key stakeholders: design, product, and data science.

Quantitative recap of the concept testing results.

Results showed that while Versions B and C were the most preferred from an experience perspective, the content in Version A was already enough to get people thinking about upgrading their plan. As a result, we continued iterating in the direction of Version A for a first release since it had high impact but required much less development effort.

The Iterations Continue

The Plan Upgrade Recommendations feature leverages AI/ML models designed by Comcast Business’s data science team to determine when a recommendation should be shown to the customer. This meant that numerous variables (organized into 4 main categories) were used to help accurately signal when a customer starts struggling with connectivity issues.

I needed to translate these findings into human language, which meant creating business rules for the copy options and regularly touching base with the content team to ensure legal compliance.

I ended up creating a content template to standardize wording for the contextualized recommendations.

The Final Design

Once copy could be set, it was time to finish up visual updates and design for breakpoints.

We always design the business internet page to account for three breakpoints: desktop, tablet, and mobile.

Reflections

I would’ve loved to have personally executed follow-up research on how well this was received by our page’s customer base before I left the team. Despite that, I have since learned that this recommendation is still used today, years later, and that further work has been done since this to incorporate additional data visualizations and ML-driven recommendations into this page.

Back to Top

Michelle Q Wang