voice suggestions on a TV screen

Voice is one of the most popular features on Comcast platforms, and a proven mechanism for driving engagement and increasing retention with customers.

The goal of this project was to further increase user engagement by improving voice education and reach both current and prospective users of the voice remote. This was done through the implementation of well-timed, contextual, unintrusive voice suggestions.

Client: Comast, Xfinty

Project date: 2018

Role: UX Lead

Partners: Michael Lacek, UI | Amanda Olsovsky, ACD

The Challenge

  • voice tips on X1
  • average monthly voice users
  • heuristic evaluation

When presented with this challenge, the only form of voice education available on the X1 platform was through a carousel menu of voice suggestions that were only accessible through a voice command such as “What can I say” or from a tile located deep within the help menu. This implementation lacks discoverability, was limited in its capabilities and flexibility as an image-based carousel, and most importantly did not provide the kind of contextual support required to convert casual users into habitual ones.

In order to meet users in the moment I first had to identify who those users are and how they do or do not use voice on the platform. Through voice data I was able to break our user base into three basic categories: light (≥ 16 commands/month), moderate (≥ 100 commands/month) and heavy voice users (≤ 100 commands/month).

Light to moderate users became the focus as they totalled about 90% of all voice customers. I further broke this data down to the kind of content these groups search for with voice, those are: direct tune to a live channel, title search and app launch. Within these categories it made the most sense to target more passive moments—such as title search and app launches—than more active moments (tune to live channel) as a way to catch the user in a more sit-back and browse moment where they might be more receptive to a minor disruption than when tuning to a live event.

The Solution

  • brainstorming
  • sketches
  • sketches
  • web content brainstorm

For implementation I teamed up with a visual designer to work out the on-screen feedback and UI, and a content writer and research partner to begin crafting the structure of the voice suggestions.

With the visual designer we determined the most obvious location for the voice suggestions should be in the same space as the pre-existing voice feedback (located in the top-left of the screen). This was already an area that voice-users looked to in order to confirm voice input so it felt like a natural place to continue the conversation by way of a voice suggestion. Looking to draw attention to the suggestion to the user we added a subtle animation to how the suggestions were presented on screen. We also added a bold + italic typeface not used anywhere else within the platform to further emphasize the specific voice command within the string of text.

Since the focus for suggestions was on title search (such as ‘Game of Thrones’) or an app launch (a such as saying ‘Netflix’) I created a list of the kinds of things we should be suggesting in those moments to offer guidance to the content designer for creation of the suggestion copy. Together we crafted a series of variations on different ways to structure the sentences for each suggestion, did some brief user testing with support from our research team, and determined that the most successful commands would occur at the END of the line of copy, and the shorter the command the more likely a user would be to use it.

The Aftermath

  • home
  • business expanded
  • company home
  • profile

The combined effort of the team has lead to the creation of 100s of voice suggestions throughout Comcast platforms (beyond the initial launch on X1). Since the launch of voice suggestions voice engagement has consistently increased year after year, the immediate impact was a 45% increase in voice commands with a steady climb of about 5% increase year-after-year.

View More...