Personalisation has become a buzzword in the news media industry. But what does it mean — especially for for news media companies moving forward?
During the final session of the INMA World Congress of News Media, sponsored by Ring Publishing, INMA initiative leads Jodie Hopperton (Product Initiative), Ariane Bernard (Smart Data Initiative), and Greg Piechota (Readers First Initiative) discussed personalisation and how it affects each of their areas of expertise.
Hopperton started by defining personalisation, noting that it can sometimes be confused with automation. But while automation relies on a set of rules to deliver content, personalisation “is really about creating content or giving content to people based on their behaviours or their desires.”
The purpose of personalisation is rooted in the understanding that readers are different and, therefore, have different reading interests. Personalisation is really about creating content or giving content to people based on their behaviours or their desires, Hopperton said: “There’s so many different needs and wants that we’ve all got, so how can we possibly serve content that appeals to everyone? That’s the challenge.”
And even when newspapers are serving up the content a reader wants, there are challenges in making sure the reader sees it. Stories get drowned out by breaking news, sometimes the reader visits the homepage at the wrong time, or sometimes they just might miss something. By personalising content, newspapers can ensure that each reader is seeing what appeals to them.
“It’s also becoming the industry standard,” Hopperton said, pointing to platforms ranging from Netflix and Spotify to Instagram and TikTok. “News is the last place that doesn’t do this. So we may need to think about what’s the cost of not personalising?”
When editors meet algorithms
The relationship between editors and algorithms often doesn’t run smoothly, and it’s a major sticking point for newsrooms looking to implement personalisation. While editors have long been considered the most reliable source for content selection, the addition of algorithms has started changing that, Hopperton said. Both approaches to content selection have merits and downfalls.
For example, an editor may think a story is important for the homepage, but a reader might not agree. And the algorithms won’t know exceptions such as major sporting events that might appeal to readers who typically don’t consume sports content. And what prevents a reader’s most-read sections from appearing on the homepage if it is soft content that isn’t suited for that page and doesn’t reflect the brand appropriately?
“We’re a long way off from the perfect algorithm,” Hopperton acknowledged. “There’s still a lot of different arguments and a bunch of things to consider.”
That includes looking at how personalisation works together with editorial and the highly revered page one meeting for big brands. How do publishers ensure they maintain the institutional/brand integrity? In some cases, those questions are still looking for answers.
But Bernard, who spent time at both Taboola and The New York Times, said those questions are worth looking into.
News publishers who jump into personalisation today have more out-of-the-box solutions available, Bernard said, which means they don’t have to build their own platform. Instead, they can access more solutions at a lower barrier to entry. She added the publications that will find the most benefit from personalisation are those with a diverse audience looking for different types of content.
Piechota looked at personalisation from the standpoint of what will be top of mind for CEOs:
“I think the CEO will be looking at return on investment. [They] will be thinking about how actually using personalisation techniques will influence KPIs across the funnel or whatever model you use to describe how you get customers to fulfill their goals with our business and across the value chain.”
Personalisation could be the answer to keep people coming back, but tracking its effectiveness could be a challenge, Bernard said.
“Personalisation itself doesn’t have its own sort of final KPIs,” she said. “All that personalisation helps you do is basically improve on those numbers. The algorithm still needs to basically track the way that it would know it did better, which is usually clicks, but your KPIs don’t change for personalisation.”
She said A/B testing, such as following one reader on a personalised experience and comparing it to the journey of someone not using personalisation, is one option — but keeping all things equal to ensure accurate results could be challenging.
However, the challenge could be worthwhile: Piechota added that a recent research paper from Netflix outlining its approach to its recommender system showed personalisation dramatically reduced churn.
“Netflix calculated that basically because of their recommender system, the impact on churn is around US$1 billion a year for them,” he said. “So this is a significant improvement.”
When to personalise
With such potential financial benefits, and the cost to implement falling, shouldn’t all publishers go all in on personalisation?
Piechota noted there are still risks and suggested that collaborating with other publishers — such as seen with the DRIVE initiative in Germany — could help mitigate the costs. And there are other considerations, one of the biggest ones being the newsroom.
Bernard explained that newsrooms must have confidence that rules that will prevent certain content (“like a bunch of cat videos”) from being optimised and landing on the homepage.
“That can give confidence to an editor who cares deeply about the expression of the newsroom’s choices,” she said.
Piechota echoed the need for editorial to be comfortable with the prospect of personalisation: “What editors want is to have a stop button so if something goes wrong, they [can] change it. If you give it to them, then you can go very far with automation and personalisation.”