By Dexter Lin
of Guardian News & Media in Australia, Dr. Deb Goswami
of Singapore Press Holdings and Saemmool Lee
of Dong-A Ilbo in South Korea shared case studies on how their media organisations have built their data operations. The panel, held at Digital Media Asia 2021, was moderated by Rosi Doviverata
, Managing Editor – Digital and Readership Development of the Fiji Sun.
Ethics drives data. That’s the view of Sandra Moynihan, who is Head of Product & Platforms at the Australian iteration of The Guardian.
“Using data with ethics and principles is really what the data approach for the Guardian is built upon. We look at everything through ethics – that’s our journalism,” Moynihan says.
Transparency and security
Transparency and security is a core value of the Guardian’s data operations. To that end, Guardian Australia makes sure they declare all their analytics partners, which says partners are approved through an internal governance council, and that they are IAB-compliant.
Any potential analytics partner goes through a thorough process of testing lasting 12-16 months, according to Moynihan.
To build trust with readers, advertising and targeting must also be transparent. For instance, advertisers and other partners are unable to use the data of Guardian’s readers for profiling, segmentation or retargeting, without prior agreement.
“We always want to be able to let [our partners, clients, advertisers and customers] know what we are doing with their data,” says Moynihan.
Good communication is essential
The increasing pace of change in technology means that communication between the Guardian and various stakeholders is key.
“Technology changes quickly. We always have to stipulate what happens if there is a change, and I think that’s really important to be really transparent and honour each of your contracts,” Moynihan says, while citing differences in data changes due to technology in the past year.
She recommends getting audits done periodically – Guardian Australia does one every six months.
She also recommends “build[ing] an in-house wiki and how to interpret your data dashboard.” This makes it easier to find what you are looking for, in the long run.
When it comes to choosing a third party vendor, Moynihan’s priority is choosing one whose ethics align with the Guardian’s: “We have to look at their principles, their policies, if they’re sustainable… we want to work with people who believe in what we believe.”
Though the wealth of data one can attain is attractive, Moynihan stresses that the endpoint is still the user. “Differentiate needs and wants,” says Moynihan. Otherwise, it may overload the user’s browser and make the reading experience worse.
Know what you want
Dr. Deb Goswami, Head of Data & Analytics at Singapore Press Holdings (SPH), also cautioned against the pursuit of data for data’s sake.
“I argue that being data-centric or data-informed, in and of itself, doesn’t really have that much value. Don’t do it unless you know what you want,” says Goswami.
He listed out several different problems one might face: moving a print audience online, a high bounce rate on articles, or augmenting advertising revenue without compromising any element of journalistic integrity – all of which require different data-based solutions.
Data is not a “silver bullet,” Goswami says.
Instead, he lists out five pillars which allowed SPH to begin the transformation into one that is data-centric.
The first pillar, which Goswami considers the most important, is culture: staff must be acquainted and comfortable with using data tools. This means in-house training, integrating data analysts and scientists into business units, and the creation of a working environment devoid of ego and politics.
“If you’d like to do something with data, don’t be intimidated by it… leverage it,” says Goswami.
The second pillar is technological.
“Pick tech that is ready to use for your organisation,” Goswami says, “don’t invest in tools if you don’t have the ability to find the outcome attached to it.”
The result will only be additional overhead, he says.
Data should serve decision-making
Similarly, data should be used as a service. If an additional dashboard will not influence your decision-making, Goswami suggests you probably will not need it. Lastly, things will get messy quickly, so investing in data lineage and governance are a must.
The third pillar is partnerships.
Goswami suggests examining what the misalignments in data use-cases are across different units in a multi-newsroom organisation.
“Which tools, and which communities that you are not engaging with?” asks Goswami.
Sharing workflows across departments within the organisation also helps speed up and lessens the workload.
Goswami also warns against “falling into the trap” of building everything internally or externally. With the latter, the problem is that the technology is not tailored to any specific organisation.
“Figure out which parts of the technology stack are absolutely essential. We have the expertise to build internally… and the rest of it we will look at engaging external systems,” says Goswami.
The fourth pillar is prioritisation.
If one is doing everything, then everything suffers. Instead, Goswami says that by using tools that you actually value, making incremental improvements becomes much easier.
Finally, the fifth pillar is adopting an iterative mindset.
Customers change, mindsets change, content evokes different responses – so when you have implemented a data-informed solution, think about how you can revisit that. Do not move on from implementations prematurely, says Goswami. Likewise, timelines must adjust to the fact that progress is incremental.
The results from SPH’s experience are encouraging. Goswami says the company now has access to predictive models, data products and A/B testing support. This has helped SPH identify users who are likely to subscribe or terminate, centralise their reach metrics and carry out their first gamification for The Straits Times.
There are, however, areas to improve as well: many dashboards have no usage, and there have been premature abandonment of experiments with sub-par results.
Korea’s news eco-system presents a unique challenge, different from the ones faced by both Moynihan and Goswami, according to Saemmool Lee, Head of Innovation Lab at Dong-A Ilbo.
Studies done by the Korea Press Foundation and Reuters Journalism Institute have found that 72 percent of Korean readers get their news from news aggregators.
Just 5 percent get news from news websites or apps. Only 13 percent of Koreans have ever held a subscription to a news outlet – lower than the global average of 18 percent.
This also means that Korean news search engines have a large amount of control over news outlets, and their data.
“[The search engines] provide a lot of data, but this data is not integrated to the data systems of news organisations,” says Lee.
To break this stranglehold that search engines have on the Korean market, Lee and her team at Dong-A Ilbo had to figure out a way to get negotiation power.
One way is to strengthen Dong-A Ilbo’s platform so that readers come straight to the website rather than through search engines. The other is to use data as a means to enhance the outlet’s journalism.
This can be achieved with data in two ways. The first is to use data to help plan quality journalism. The second is maximising distribution with the help of data. Lee’s team collects data from across the organisation and cross checks with other departments, before delivering reports to management with holistic analysis.
Data isn’t everything
Holistic analysis is key: Lee cautions against using data as the sole factor for planning stories.
“Our goal is journalism, not the data itself. We are trying to make sure that our reporters provide stories better, and enhance our reputation through their stories but… [we don’t tell them] you have to produce this amount of data[points] because it sometimes can destroy the behaviour of the reporters,” says Lee.
Distribution also cannot tell the full story.
“Some stories get really high interactions, but readers might not remember who wrote it, or which application published it,” says Lee.
Another shortcoming of using data without context is the possibility of creating stories that are sensationalist. A story that may have many hits might just be due to outrage, according to Lee.
In that case, a conclusion might be drawn that in order to hit the targets set by data, journalists may be pressured to seek out outrage, or write in a way that invokes extreme emotions rather than keeping to what Lee calls the “essence of journalism.”
Lee says the same data can be interpreted differently by different stakeholders.
However, since News Innovation is independent from the newsroom, it is easier for them to spot biases from within.
One of the results of this data-driven research is The Original – original.donga.com
. Seeing how most readers only skimmed through articles without interacting much, they started a website for ad-free, multimedia investigative journalism.
“We are still experimenting and we are still analysing the data results, and to see what we can learn about this initiative but we are trying to build a system,” Lee says.
“I really would like to have a healthy cycle of news delivery and consumption… Quality journalism, increasing audience experience and brand reputation leads to stable business. [With this], we might be able to invest more in quality journalism.”
About the author: Dexter Lin is a third-year undergraduate student at Yale-NUS, Singapore. His interests lie in the effects of Orientalism in the media and wider culture. Outside of school and work, Dexter also represents the Singaporean national team as a 100m sprinter.