News

Publishers share tips on building a data-centric organisation

From audience engagement to advertising, more media companies are looking to adopt a data-driven culture. Data-savvy publishers shared how they are using data to understand their audience in this Digital Media Asia eSummit 2020 panel moderated by Sally Wu, Director, Business & Audience Insights (Research & Analytics), at BBC News Global Singapore.

by WAN-IFRA External Contributor info@wan-ifra.org | October 20, 2020

Above image: Mediacorp’s award-winning universal ID integrates the disparate forms of data collected as users journey through the publisher’s online properties. (Slide: Sandeep Rai)

By Alysha Chandra

“A creative idea is only as strong as it is when backed by data,” said Sandeep Rai, Senior Account Director, Digital Media Solutions at Mediacorp. But the data publishers have traditionally relied on could become more scarce.

As calls for greater privacy controls grow across the globe, major web browsers like Mozilla Firefox and Apple’s Safari have banned the use of cookies, small pieces of data that help to track a user’s activity across different websites.

Google has followed suit, announcing that it will phase out third-party cookie collection on its Chrome browser, which makes up over half of the global web browser market, by 2022.

Rai shared how Singapore’s national media network is preparing for a world without cookies.

“We were very reliant on third-party platforms for data purchasing and applications. Later on, we moved into licensing a data platform to segment our own first-party data. The next step was to build our own platform,” – Sandeep Rai, Mediacorp

Mediacorp’s answer to this problem, said Rai, is MeID, the company’s proprietary audience analytics hub.

As users journey over Mediacorp’s online properties, data is collected through web analytics tools, advertising servers, recommendation engines, data management platforms, user logins and affiliate partners.

Rai said a huge advantage of MelD is the way it consolidates these disparate data sources. “That single source of truth is so important. Now I’m no longer comparing between different data platforms but viewing a single source of data,” said Rai.

Benefits from this platform have rippled across the organisation, he said, enriching content recommendations, enhancing data portability, fine-tuning ad targeting and saving costs.

“Through MeID targeting, our advertisers get a better proposition, as well,” he added. MeID has even bagged multiple awards for excellence in data at the Global BIGGIES 2019.

Rai advised other publications to first consider the purpose of their content platform and audit where their business stands in building a culture of data before deciding whether to build a web analytics platform internally or whether to license third-party software.

“If you’re not at a phase to build your own platform, it is not to say that you’re completely helpless,” said Rai, suggesting that participants could look to Lotame, Iota and Google, which are working on their offerings to help publications deal with the phasing out of cookies.

Rai also spoke on the challenges of building a data culture in a traditional organisation like Mediacorp, where data-led thinking is a top-down approach.

“It is not just building up a team and ensuring that the right assets are available, but convincing everyone who works in the company and who deals, either from a commercial or content creation side, what tools are available and to think from a data perspective,” – Sandeep Rai, Mediacorp

Identifying your most valuable subscribers

“Like many publishers across the world, we are looking to measure the contribution each subscriber makes to our bottom line,” said Fiona Chan, the Head of Analytics, Customer Experience and Business Operations, Singapore Press Holdings (SPH).

SPH, which is the biggest publisher of newspapers in Singapore, has done this by measuring the Customer Lifetime Value (CLTV) of their subscribers.

“If you want to do any kind of forecasting in terms of the performance of your company, it starts from the bottom up,” said Chan.

Calculating the CLTV, which measures a customer’s current value as well as their potential future value, allows SPH to know how much revenue their customers will bring in.

This allows the company to know what costs they can afford to incur to attract this revenue, allowing them to granularly target their advertising, subscriber retention and product development spending, said Chan.

There is no one approach to calculating a CLTV, with decisions including whether a company is looking to measure revenue or profits, as well as the period of time over which the value is calculated.

“Whatever works for your organisation depends on what data is available, how many customer segments you have, what you feel comfortable targeting, and how easy it is to get that data. If it takes a long time to get that data, it gets outdated very quickly and doesn’t help,” – Fiona Chan, Singapore Press Holdings

SPH calculates their CLTV by looking at cohorts of new subscribers in each calendar month and sorting by product and package. So far, the company has found that products with print components tend to be associated with a higher lifetime value.

Chan explained that print packages tend to be priced higher than digital packages and tend to churn less. Chan also found that subscribers who take up promotions on digital products tend to have a lower CLTV as they are price-sensitive, have less stickiness and tend to churn higher.

Calculating the CLTV has had a real impact on SPH’s bottom line.

“In the two to three years since we began calculating our CLTV, we have used it to put a cap on our acquisition costs and have seen a 25% reduction in per customer acquisition costs,” said Chan.

Humanising data with user personas

The South China Morning Post has taken their customer profiling a step further, using the data they have collected to create archetypal profiles of customer segments called ‘personas’, said Korey Lee, the publication’s VP of Data.

To build these personas, SCMP’s data team aggregated a variety of data, including behavioral data from Google Analytics and demographic data collected through surveys, quizzes and polls on their website.

Lee emphasised the importance of well-designed questions.

“We might ask readers when the last time they took a vacation was, or if they are planning to bring their kids along. This tells us whether they have children or not and maybe their approximate age bracket,” – Korey Lee, SCMP

The team then leveraged machine learning to extrapolate that information across their entire user base and build data profiles based on these segments.

To bring this information to life, the company also spent time interviewing users to add a qualitative element to their profiles, said Lee.

As SCMP launched their new subscription offerings in August, they identified and named six personas with varying degrees of engagements and subscription levels, ranging from ‘lifestyle-news seekers’ to ‘inquisitive professionals’.

“Our big takeaway was that SCMP meant different things to different people,” said Lee.

Above: SCMP assembled six different consumer personas ranging across different levels of engagement and preferences. (Slide: Korey Lee)

For example, the “reserved local” persona skews older and has a larger mobile app engagement, with 77.9 percent of them having downloaded the SCMP app. Their higher engagement suggests that the company should focus on converting them to subscribers, said Lee.

“Having a different strategy for each of these personas and each of these cohorts is really important to us on how we can kind of optimise not only our conversion but engagement with readers,” – Korey Lee, SCMP

Beyond subscriptions, understanding these personas has also been useful to SCMP’s editorial, social media, product, and marketing teams.

“These personas help our editorial teams to understand who they’re writing for and our product team to design different experiences around each reader,” said Lee.

Lee also spoke on the post-cookie world and the imperative it has introduced for publishers to more meaningfully engage with users and encourage them to log in to platforms.

He suggested that publishers offer incentives like customised content and newsletters, as well as the ability to engage with editorial staff, writers or experts in a particular area.

“We can’t be too cavalier; we have to give users a cookie or a reward as well to encourage them to continue logging in,” he said.


About the author: Alysha Chandra is a third-year student at Yale-NUS College in Singapore. 

Edited by Bill Poorman.

WAN-IFRA External Contributor

info@wan-ifra.org