O Povo

O Povo Online

O Povo Online is the digital branch of O Povo Group, an 8-decade old media conglomerate in Brazil. In 2015, with the stagnation of its user engagement metrics, the arrival of new competitors, and the loss of leadership position they decided to reinvent its news portal. Soon they realized how complicated this task was.

The Challenges

Many opinions, little evidence

The project task-force (17 people from management, marketing, journalism, social media, engineering, and sales) was struggling with pre-conceived opinions about problems and potential solutions. Although some data (web analytics, surveys, etc.) were available, it was hard to turn them into useful insights. Consequently, the guesswork was the norm.

A  release-everything-at-once product culture

The tight deadline (five months to deliver a “new portal”) created an atmosphere of uncertainty given the complexity of the product. The management expectation was to deliver an entirely new experience, all at once, covering all areas of the portal.

My Role

I was hired as an external user experience consultant to facilitate the process of UX strategy redefinition and execution throughout the project. I led research activities, panels, ideation workshops, and prototyping sessions. I also conducted some usability tests and card sorting.

The Approach

The first action was to start driving decisions based on data and facts, eliminating the guesswork from the process. The second initiative was to adjust the mindset from delivering “a whole new portal” to “something meaningful for the user”, iteratively learning from small experiments instead of aiming to release a big, monolithic product.

Empathizing with users based on data and facts

Firstly, we created sub-teams to collect and generate qualitative and quantitative data from primary and secondary sources: access logs, direct user feedbacks, previous surveys, public statistics, social media, customer interviews and usability tests with existing products. The results were presented and discussed in weekly seminars, with essential user behaviors and other characteristics being identified.

The next step was to create user personas from the data and insights generated from the research. To find patterns and user segmentation, we used a clustered matrix (behavior x attitude x demographics) initiated on a board and consolidated in a spreadsheet. We found two main clusters that were turned into personas.

Segmentation table
Segmentation table

Finally, we used empathy maps to explore the personas and framing their most significant problems. The team was split into two groups, one for each primary persona (David and Ana).

Empathy maps creation
Empathy maps creation

 

Persona Ana
Persona Ana: the non-stop information consumer (empathy map representation done by the team)

Ideating

In the ideation phase, the team generated around 40 potential solutions for the problems found previously. To organize and prioritize it, we use an Impact (benefit) x Viability (technology and content) matrix, divided into four quadrants, suggesting the order of execution.

Ideas matrix
Ideas matrix

Next, we grouped similar or correlated ideas and turned it into small initiatives (experiments) in a unified list of 7 projects with two lines of action: Redesigns and new offerings. The concept of experimentation was extensively stressed, meaning there would be no fail, just learning within a controlled environment.

Experimenting

The first experiment was a new offering: a beta version of a content delivery service via WhatsApp. The results were monitored using a spreadsheet. We tested several content types, frequencies and sending times, with conversions measured daily.

WhatsApp News Service
WhatsApp News Service

 

The other experiments took place on the redesign side. First, the team focused on new contents, aesthetics, and ease of use for the mobile portal—identified as a high priority based on the user research findings. The desktop version came next, with a redesign of the News section only (the idea of redesigning the whole portal was abandoned during the ideation phase).

For the content reorganization, we ran card sorting studies (one for each persona) to validate the categorization ideas.

Card sorting analysis (Similarity Matrix)
Card sorting analysis (Similarity Matrix)

 

For the prototyping, we ran design studio sessions with the whole team (designers, journalists, developers, etc.) collaborating.

Design studio session
Design studio session

 

For validation, we used a range of techniques, including 1) rapid usability tests on paper, 2) A/B tests in UsabilityHub (for typography) and 3) recorded usability tests sessions using Camtasia and Lookback.io.

Usability Testing (screen capture)
Usability Testing (screen capture with the participant’s face on the bottom-right)

In parallel, we intensified the social media initiatives with two lines of action: 1) creating new types of content (mostly video based) and 2) adjusting the language to a more lightweight tone, according to the persona profile.

The Results

Boost in engagement and audience
Just three months after the first release and several weekly tweaks, the mobile version had an increase of 245% in the unique visitors.

Successful new offerings
The experiments with WhatsApp evolved into a new product, growing exponentially and receiving excellent feedback, becoming a benchmark for similar companies.

More advertisers
New advertisers arrived (figures not disclosed).

Change in the internal mindset
This project changed the product development culture profoundly, according to the team, which now embraces failure and experimentation as part of the delivering process.

Leaders again
Finally, they reclaimed the lead position in the local audience (portal and social media). They even made a t-shirt to celebrate!

Celebration t-shirt
Celebration t-shirt: “1 Million likes on Facebook, 500k followers on Twitter, 200k followers on Instagram. We are O Povo Online, Audience leader in Ceará”. Note: All figures above have increased significantly since then.

Learnings

Focus on data/evidence is an effective way to avoid excessive guesswork in design without turbulence among team members.

Small experiments that generate quick results yields confidence and buy-in from both management and product teams.