DEEP-DIVE CASE STUDY

Twitter TV Ratings

USER RESEARCH | STRATEGY | INTERACTION DESIGN | VISUAL DESIGN

Every day on Twitter, hundreds of thousands of people tweet about TV shows as they're aired. Major networks, like BBC, use key tweet metrics to determine their network's overall popularity on Twitter. These metrics include: number of tweets, overall engagement and sentiment about specific episodes. We set out to build a custom TV analytics platform that would answer key questions about TV engagement on Twitter for a variety of partners around the world.

Meet Sarah, a TV Analyst at BBC

  • Sarah starts every morning the same way: She logs in to Twitter TV Ratings to see where the BBC's shows fall on the leaderboard for most tweets, most authors and most impressions during the episode's telecast.
  • At a moment's notice, Sarah needs to be prepared to report on any variety of show metrics (total number of tweets, number of unique authors, overall tweet impressions, number of tweets per minute and demographics) for both BBC and competitors' shows— this could be delivered to show producers, advertisers or network executives.
  • On a monthly basis, Sarah prepares in-depth presentations showing the projected engagement numbers for the following month, as well as any emerging trends in overall engagement.
  • For network premieres or special events, Sarah sends out a report to everyone at Sky showing a timeline over the course of the telecast including— tweets per minute, overall tweet volume, sample tweet content, and demographics (male vs female, age ranges and locations).

Daily Reporting

Sarah uses the main dashboard (Leaderboard) to see how last night's shows ranked and get a quick glance at a sample tweet.

Special Reports

The telecast details page allows Sarah to quickly create a report showing overall tweet volume, average tweets per minute and most engaging moments from the show (note: wireframe and sample report do now match)

tweetable1b.gif

Project Approach

Sarah hates spreadsheets

To develop Sarah's persona, we talked with a number of analysts from networks, cable companies and internal Twitter TV teams to gain a better understanding of their process and how they use Twitter data in their current workflow. Before our product launch, their world was bleak— when I asked one of our data analysts for Germany's most talked about TV shows during a two-week period in April 2014, he sent me a total of 14 spreadsheets, each with 55-rows of TV shows, spanning 18-columns of data. All of this, just to see which shows had the highest tweet volume and the biggest reach on any given day. 

Understanding Tweet patterns

Just to gain a bit more context for how shows are talked about, we sat down with our internal data team to look at how tweets and engagement changed depending on show's genre. We needed a better understanding of the data so we could highlight the most relevant tweet information, at the right time. As shown below, you can see that films are a big driver of social engagement and tweet volumes stay elevated throughout the entire event, with small peaks surrounding iconic scenes. Competitions on the other hand, yield a fewer number of overall tweets, however extreme peaks appear during performances or head-to-head competition. We used this information to help inform how we would display tweet volume over a 2-3 hour period of time.  

Object-Oriented UX / Short-hand Wireframes

With a team spread out across multiple timezones/locations, and a limited amount of time before launch,  we found it necessary to wireframe early and often, as we collaborated with our data analysts and developers to insure that we were delivering the right data at the right time. As you can see below, we developed a bit of a wireframe short-hand with stylized sketches focussing more on data types and attributes, and less on design. This allowed for our development team to build an entire API backbone without waiting for completed designs. For those of you following along at home, you're probably wondering why we didn't use the existing Nielsen APIs. Unfortunately that was built on an outdated platform, so we had to start from scratch, building our new product into the Twitter API pipeline, while offering flexibility of content and data to different partners.

Testing + Iterating

Shown below, you can see early, medium-fidelity wireframes that we tested with a handful of our customers. During this phase, we were experimenting with a "score card" at the top of the screen showing key metrics from a specific episode and then two tabs, one for Tweet volume and the other for User Insights.

Flexible layout for partner sites

In addition to creating a Twitter-branded TV Ratings product, we wanted to provide our partners, such as Gfk, with an internal dashboard that meshed with their existing TV metrics dashboards. Together we built a 12-grid structure that allows for flexibility while also driving design decisions.

Results

Based on our customers' success using our TV Ratings tools, we solidified international partnership deals with Gfk, Kantar and Video Research in 2014. With these partnerships, Twitter TV Ratings are now available in North & South America, most of Europe, South Africa and Asia. As shown below, we modified our Twitter TV Ratings dashboard to match Gfk's brand guidelines, and began to flesh out a flexible grid structure to allow custom integration with other partners.