Case Study: Ge — Broadcasting live content on a massive news portal

Broadcasting live content on a massive news portal.

Ge, or Globo Esporte, is LATAM's top sports platform, providing live matches, scores, and compelling editorial content.

Sports

Broadcast

Web

iOS

Android

tvOS

B2C

DTC

Before we dive in, let's set the stage about Football and Globo.

You know Brazilians love football, right?

Of course you do. Some people love it so much that they don't miss a thing about their teams. In fact, 1 every 4 Brazilians access ge every week.

This devotion is rooted in Globo's rich tradition of broadcasting live soccer on TV in Brazil. Over the past decades, Globo's broadcasts have been a fixture in more than half of Brazilian households during evenings and weekends.

Established in 2005, ge emerged as the premier online hub for football news coverage, bridging TV content to the internet. After fourteen years, the dedicated team behind the product embraced a new mission.

In 2019 Globo decided to change ge's business strategy.

Relying on free articles and 15-second ad-supported VOD videos became unsustainable over time. Competitors were causing a considerable impact on the streaming business, prompting Globo to act.

To challenge high-level competitors, ge began live streaming matches — an unprecedented step for the company. However, this posed significant technical challenges, especially given the bustling nature of our homepage.

💡 Insight

How could ge establish itself as the go-to for live content streaming in Brazil?

How did we translate this business goal into a valuable feature for our users?

With this question as our guiding principle, the team began brainstorming ways to design a minimum viable experiment for testing video content on the homepage.

”Our goal was to not only start offering a better match experience to the audience, but to set GE as the go-to place for watching it too.”

Moisés Ribeiro

Project Manager on this project

Given its high traffic, the homepage was the perfect place for our experiment. If successful, this could create a new revenue model, with potential for expansion and improvements.

Understanding matches

In Brazil, football championships often have simultaneous matches.

Thus, the homepage functioned as a match aggregator, summarizing games through highlighted scores.

Drawbacks of the previous user interface

The previous highlight component was designed exclusively for supporting hard news format.

It failed to capture the match excitement and lacked effectiveness in driving new subscriptions. Additionally, there needed to be more consistency in displaying content on web and mobile platforms.

First validation

After several stakeholder interviews, the team felt confident to advance the design testing.

“We wanted to learn if people could use the new components and their perceptions. Also, if people's perception of live content had improved - and if there is anything we should look for before making it a live experiment.”

Marcela Maués

Lead Researcher on this project

We screened and invited 20 users to our Globo headquarters for prototype testing on both desktop and mobile apps. These participants were ge users, encompassing diverse profile types and usage patterns.

As confidence in the tests grew, it was time to take the experiment to production.

We shifted priorities to refining the design. The aim was to develop a simple, scalable solution for easy implementation.

Validation & rollout

The validation phase started gradually. Initially rolling out to a small percentage of the user base, exclusively during live match timeframes.

We provided the editorial team with best practices to highlight matches within during the tests.

+35,6%

increase in match page sessions

+6,9%

increase of live video consumption

12 days of individual tests

Working closely with the editorial team, we curated various match offerings to ensure a diverse and comprehensive dataset.

Time-bound A/B tests

Conducted within pre-match, live match, and post-match periods, approximately 5 hours each day. This approach provided isolated metrics for fair cross-comparisons in the future.

To confirm whether users perceived any difference, we designed a custom NPS interface to promote the rating experience.

By building a personalized and native NPS interface, we aimed to offer a smooth interaction to those wanting to give feedback.

13.2

NPS Score on the
older version

22.5

NPS Score on the
new version

Throughout the A/B testing phase, both versions of the homepage featured the NPS interface, enabling us to compare scores between the two designs

Outcomes & learnings

Although the results were positive overall, the experiment couldn't maintain advertising impression rates. This posed a serious threat, as advertising was the primary source of product monetization at the time

In the end, we managed to maintain the same rate of advertising impressions as the previous homepage without making significant changes to the new designs. Below is a chart illustrating different timing variations for displaying advertising positions.