For UOL, the Meta/INMA Accelerator project was an opportunity to determine which content to close behind a paywall. Project leader Emiliaine Vieira said the company already had a paywall but wanted to expand its subscription product, which currently is limited to just a few areas of its site.
UOL is owned by Grupo Folha and stands for Online Universe — the largest content, tech, services, and digital payment media company in Brazil.
The company has some specific rules it follows, including not being able to paywall the following content:
- Gate traffic originating from the home page.
- Hard news content.
- Non-opinion columnists.
- Merging open and closed content in the same section.
A new paywall experiment
UOL decided to work with Piano’s paywall in the future so it can work within the confines of its business rules. But it also conducted a series of analyses and defined areas to test the implementation of a paywall.
“Our first goal was to learn to know our subscribers better,” Vieira said. “We studied their behaviour and demographics. We look at what they consume across the site, beyond the sections, and we compare that to the general users to compare how similar they are and how different they are.”
The next step was for UOL to have a better understanding of its products, such as the sections with the highest conversion rates. It extracted entities from the most-read content using NLP (natural language processing).
Then UOL created these new paywall rules to test:
- Apply hard paywall to content that already has a metered paywall.
- Apply hard paywall to content on Google AMP that already has a metered paywall.
- Paywall the open content that is most often read by subscribers.
- Paywall content to those with the highest conversion rate.
“We’ve identified 100 tags and four traffic sources that are eligible to close with these rules,” Vieira said.
Next, UOL plans to run experiments to validate the rules it identified and repeat the process to identify even more rules and opportunities. Long-term, it wants to use Piano and machine learning to move to a dynamic model based on the user journey.
Vieira’s advice to other media companies is to cross-reference data on Google Analytics by user ID to see what users are doing across multiple sections. She said they were excited to implement everything they learned all at once, but they quickly learned to take it one step at a time.
“We need to start with the basics of our business first,” Vieira said. She also learned simple and clear data is more useful than complex data that has no viable use. And most importantly, she learned a model based on the audience is best for starting out; from there, you can evolve if you know your users and their journey.
This case study originally appeared in the INMA report, The Benefits and Risks of Media Data Democratisation.