When it comes to the data ecosystem, finding what you need from it and tying it to business objectives may be more difficult than implementing it.

At the INMA/Meta Audience Analytics Town Hall on Tuesday, three media executives share how prioritising data strategies increased reader and advertiser revenue at their companies.

Media24, South Africa

Gareth Lloyd, head of research and analytics at South Africa’s Media 24, shared some of the learnings from his company’s digital journey, which began in 2016.

The system they inherited at that time was an ad hoc collection of platforms and systems that were mostly built in-house: “The best way to kind of characterise it was that it was held together with sticky tape and bubblegum. And there were a lot of manual steps in between.” But, he acknowledged, “It fit our maturity at the time.”

However, as they began wanting more from data, the company pivoted and built a new subscription platform that became one of its key components.

“This gave us the perfect opportunity to say our brand is going to be completely replatformed. We need to think about completely redoing this in terms of what we have access to from a data ecosystem point of view.”

That included looking at things like how the product works, what to start outsourcing, what to report on, where to host the data, who should do what — and what they should be doing. That journey is not yet complete, but Lloyd shared some key insights gained along the way.

One of the biggest changes was in how they thought about the behaviour of customers. They moved to an ecosystem view, focusing on the user-level data and stitching it together across various data source components.

“That is where the data strategy actually started to feel like the company’s strategy, which was eventually going toward a customer or orientation,” he said. “We completely reoriented the focus of the company towards the customer.”

User data was a key component in being able to make that shift, as it allowed them to create reports and analyses about users and retool operations in accordance with customer behaviour. Data allowed them to observe the interactions with customers across all touchpoints in the ecosystem and identify how often they used certain products.

From there, they could group customers and sell advertising differently: Instead of selling to a general audience, the ad could be targeted and served to a specific group of users.

“So when the user came back to the site, regardless of what brand they landed on, they would be served ads relevant to their interests based on their prior behavior and the segments they fell in,” Lloyd explained. That also affected editorial approaches, as they could understand the differences between subgroups and serve them content that would best engage them.

Understanding the phases of this five-year journey began with identifying the North Star metrics and then looking at what they needed.

“Remember, we were starting from basically nothing, no data ecosystem. We were mapping out our sources [and] what we needed. We wanted segmentation. We want dashboarding. We said we wanted predictions, but we didn’t use that until much later on,” Lloyd said. “Once you've identified the data sources that we need, we started to understand ecosystem synergies.”

Something to take into consideration is the ease of implementing each one. “So for example, where there were products that allow you to almost plug and play based on some of our existing set-up, we would go for those because it was effectively the low hanging fruit.”

Setting up a data ecosystem should be done “one bite at a time,” he said, and offering six points for listeners to remember:

1. Structure follows strategy. Be clear about what you want to measure and how it maps to your strategy.
2. Collect customer-level data. This is the only way to be truly customer-centric, so design your ecosystem around this principle.
3. Establish your MVP (minimum viable product). Make sure everyone agrees on and understands what this should be.
4. Outsource and stitch. Outsource the most complex jobs outside your core competency and use your data team to fetch, store, and stitch data together.
5. Get fancy later. Put your focus on getting the data right and add other features as needed.
6. Use plumbers and interpreters. Data engineers are the plumbers; use them to create your ecosystem. Then use interpreters to help create your data culture. 

Grupo OPSA, Honduras

Leonidas Mejia, IT manager at Grupo OPSA in Honduras, discussed its project on Data Management Platform (DMP) implementation and sales go-to-market strategy, which focused specifically on advertising. 

This project aimed to solve some challenges Grupo OPSA faced:

“We were leaving money on the table because we don’t have a value added proposition for our advertisers that can better target potential customers,” Mejia said. “Right now, you should move from selling content and selling audiences at volume, to selling  audiences depending on interest and behaviour.

“Implementing a data management platform could give our commercial departments a new edge with a new pitch, new sales materials based on data. This is a first step on a first-party strategy that we’re working on.”

There were three main objectives to this project: 

1. First-party data.

2. Better targeting capabilities.

3. New value added proposition.

“We needed to develop first-party data segmentation capabilities of Grupo Opsa audiences on starting from our websites,” Mejia said. “I think we have enough first-party data to leverage on.”

Regarding better targeting capabilities, Mejia talked about how the data management platform can enhance the sales value to its audience: “Our DMP could send sales propositions to clients. Advertisers prefer Google, Facebook because of their audience reach, target capabilities, self-serving options, and lower CPMs. This project will help us justify higher CPMs, offering value added propositions to our advertisers, and differentiate our brand from other competitors.”

Mejia mentioned the platform Grupo Opsa used for the data management project and how important this decision was: “This was a business initiative, not a tech project.”

The segmentation capabilities are Grupo Opsa now offers are:

• Technographics: tech, device, geolocation.

• Engagement: user behaviour, which users read full articles versus a snippet.

• Interest: based on content.

• Intent: based on clicks to campaigns.

The classification of content is a very important part to gathering data. And for the first time, Grupo Opsa used Artificial Intelligence to start automatically classifying its content based on language used in the content. “This will give us an edge,” Mejia said.

With content and advertising being classified into categories, now Grupo Opsa can build its Phase 1 Taxonomy — the cohorts and segments, which will be used to better target the advertising. 

Mejia gave a glimpse of the new dashboard that has been implemented into the data management platform. He showed a snapshot of the Automotive classified audiences, which Mejia said “shows volume, the sessions, the pageviews, the number of articles they consume, their tech data. We are already having discussions, business questions and opportunities that was impossible to have on our mind before the project.”

Mejia shared four major lessons learned during this project.

1. Stake ownership: Commercial leaders must own the project, and the vendor needs to speak the “sales language,” he said. “This is not only buying the best platforms you can get and put it to work. If no one helps you, if the vendor just focuses on tech, that will not work.”

2. Commercial and newsroom alignment: “Now we are seeing that having better alignment between the areas where commercial requests content to the newsroom based on data.”

3. Artificial Intelligence replaced manual-only categorisation: “Our heads were exploding. We were trying to nail down all the keywords and all the possibilities that we should be able to come up with to classify the text. This will help us a lot.”

4. Audience value: “This will help us a lot when we go to market our audience valued marketing, and we are going to differentiate against other publishers and platforms.”

El Comercio, Peru

Guilianna Carranza, data and analytics manager at El Comercio in Peru, detailed a project conducted by its sports news media brand, Depor. The project created the company’s playbook for building niche audiences in the hispanic market. 

The company sought to answer this question, she said: “Is the content better or equally consumed by people in other countries than Peru?” To do so, El Comercio aimed to find out if sports news is being consumed more or equally compared to Peru audiences. 

“We also wanted to confirm, or not, the globalisation of the Depor brand in new countries,” Carranza said. This was done by determining the best recommendations for Depor content that gives the highest revenue. 

Another important piece to the project was measuring key performance indicators (KPIs), which allow El Comercio to monitor all revenue at Depor from each country — essentially, to see which countries bring in the most revenue. 

There were two main hypotheses El Comercio came up with for this project. “Which is the real content interest to audiences outside of Peru?” Carranza said. This led to the hypothesis that core sports topics (tags) underperform with audiences outside of Peru. 

The second hypothesis was that targeting niche audiences would bring higher revenue for each country. 

“We started by carrying out Depor’s data journey, that is understanding how the evolution, performance of the readership visits have been in the last two years,” Carranza said. “Then, we establish those questions related to content and business that we needed to solve.”

The next step in the set up was to analyse how the traffic to the site does versus the amount of revenue coming in: “Traffic analysis, pageviews, versus our [prices], were important both in Peru and in the countries where we have readership, hispanic market. Specifically, how Depor was doing in Mexico and Columbia, in which we have already deployed the brand.

“Finally, we run the scenarios and review the results in order in order to summarise and consolidate them.”

The initiative relied on the internal team, collaborative work, data, and consultants. 

“Our first hypothesis was confirmed that the content monetized the most is not sports,” Carranza said. The data came back that the most consumed content is viral tags, not sports. 

“Our second hypothesis regarding reliant on commercial categories to be able to measure, or segment and profile an easy eCPM (effective cost per thousand impressions) behaviours was also validated given the results of the analyses and scenarios,” she said. 

So what was learned from this project? “New perspectives outside of Peru are now known, and mainly having had the opportunity to link the ‘Data Value Cycle’, that is: content, audience, traffic and advertising,” Carranza said. 

El Comercio will continue to monitor these results to have control and consistency and also continuously improve. 

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .