Gens Z and AI top of mind at INMA Media Innovation Week

Some themes, a case study and some tidbits from INMA’s Media Innovation Week in Antwerp, held in the last week of September.

Moving from “digital transformation” to “the transformation of digital”

“2023 is going to go down as a significant media turning point,” said Robert Whitehead of INMA. “We are standing right in the middle of it.” It’s hard to disagree.

We are moving from the period when print operations were transformed into digital operations to one where the digital operations themselves are being transformed.

This is not just about Gen AI, though that plays a very (very, very) significant role, but about a period where what role traditional publications play in people’s lives is under scrutiny (eg why is there so much news avoidance?).

Publishers are looking into “performance journalism”, which addresses what user need each piece of journalism serves and how it advances the business model. As Whitehead pointed out, “it only took us 200 years to work out what valuable journalism was”. 

This led to talk about the concept of “legacy digital”, which could encompass things like an obsession with page views and traffic, for example. But also actual roles: “People who want to spend all day moving stories around websites? That’s for bots.”

The consensus was that there will be increasing convergence between media types. So CNN and the BBC could end up being as much rivals for former newspapers as their competitors in the same sector. This already happens of course but as video and audio become more prevalent so it will increase.

What to do in this age of disruption

Greg Piechota,, INMA’s researcher-in-residence, gave a typically punchy talk about how to survive in this era when Gen AI threatens to remove the source of a third of traffic and a third of subscriptions for publishers. He described it as “a tremendous disruption”. 

He identified two main ways to respond:

1 “Reassemble discovery and access” – in short, grow direct traffic and your own channels; differentiate your brand; create non-text content (which is harder for Gen AI to produce … for now); improve UX using AI.

2 “Adapt the business model” – in short, focus on reader revenues; collaborate on bundled products, maybe even industry wide; ask big tech for payments.

In the meantime, the business has to reduce costs (but be aware that the more we move to digital, as distribution costs go down, a higher percentage of costs is taken up by the newsroom); ensure its products are differentiated; focus on niches.

The good news is that personalisation allows serving niches at scale. We might be moving from the era of mass media to personal news media.

He said more that I’ve included in the tidbits below, and he also believes that there is a big role for humans in verifying the content produced by bots (I agree … for now).

Gen AI – it's not just coming, it's here

No highfalutin conversations here about algo bias and wishful thinking that “everyone wants a human touch…” It was all about the practicalities and how to conceptualise AI. 

Some said we “must treat it like a co-pilot”; others that it should be regarded as an intern and given simple tasks first before letting it loose on bigger projects.

As for applications, article summaries are being done almost everywhere. All reported that it increased engagement on articles, rather than decreased it, as one might presume. At VG in Norway, they talked to readers who said it encouraged them to read more as it broke down the forbidding nature of subjects with which they were not familiar. 

A data science interlude: some newsrooms are using Chat GPT to check the accuracy of their Chat GPT-produced summaries. Pretty smart.

VG from Norway explained what editorial tools they are experimenting with: 

1 one that suggests follow-ups to stories after being fed the original article’s url;
2 another that ingests a press release and suggests an angle;
3 one that transcribes podcasts and looks for news lines within them.

Russmedia, an Austrian publisher, is also doing summaries (like VG they are checked by journalists before publication), and has initiatives around automating legacy processes, digital storytelling, SEO optimisation, and smart paywall strategies. They include:

Comment moderation
Tagging
Extending and shortening text
Headline optimisation
Summaries
Quizzes
Turning UGC into articles, videos etc
Automated content enrichment (eg adding hyperlinks)

It’s a revolution!

Fun fact: there are already 450 news sites that are fully automated.

Gen Z – fighting news avoidance

Everyone was obsessed with how you reach this tricky audience who avoid news because it’s “too negative”, “not relevant to me”, or has a “lack of authenticity”.

Now I don’t know about you but I feel this applies to me as much as it does my teenagers. Maybe in working out how to serve Gen Z, and not just on bloody TikTok, we will work out how to better serve all audiences. 

“Can a single product satisfy Boomers through to Zoomers?” was a question asked on stage. People doubted it, but I think it’s possible.

One speaker said the key to serving Gen Z was to “go niche or go personalised”. 

The niche suggestion was based on Gen Z being 44% more likely to pay for niche content than older generations – ie they know how to curate their news experience. 

Personalisation is based on the observation that they have been training algorithms for years and instinctively understand how to get the best out of products. 

And there perhaps you have the recipe for a successful news product: niche and personalised (which fits in with Greg Piechota’s observations above).

Berlingske describes its radical shift to digital-first

Mette Ostergaard, the editor of Berlingske in Copenhagen gave a fascinating talk about her publication's shift to true digital-first publishing. 

Five years ago she decided to address the challenge that “we are digital first … when the paper is done”. So she outsourced the printed paper, moving it to a building 3km away.

She says the main reason “was to change the culture – the kings and queens were now the people doing the website and video and audio”. There were two people who made sure the prioritisation of the print running order was right, but they were sat in a corner and not allowed to talk to journalists. “Symbols matter,” she explained.

The Page One  meeting disappeared, though she does give it a minute’s thought when asked. Instead they talk about what they are doing for “prime time” the next morning. 

They focus their publishing around four prime times a day during the week and two at the weekend. Their planning is not done on a Sun-Mon basis but Wed-Wed so they don’t run out of things for the weekend – because they’ve found that Monday morning is their No 1 time for attention, instead of Sundays as they had previously assumed in the print world. 

Key factors in assessing stories for the content plan are 1 relevance and 2 identification (ie, “about someone not something”). “We are doing breaking news differently too, as it’s not worth much without translation [ie context].”

The punchline: print readership grew.

Tidbits

Mediahuis is the largest seller of e-bikes in Belgium!

As an experiment Greg Piechota used AI to recreate development of FT Edit, nominally reducing development time by 50%. And to reimagine Guardian premium offering: it came up with options to read, watch or listen to any article

A German regional publisher (NWZ) has a podcast written, voiced and produced entirely by AI.

There was a general feeling that subs are not remotely near their peak, but bundles are the new trend. Roularta, a pan-European magazine publisher based in Belgium, has an upsell for its individual magazines that allows subscribers to access other titles. Particularly popular with women who want a women’s mag alongside their current affairs staple (Knack).

Toronto Star gets 80% of its traffic from 20% of its stories, and 80% of that traffic comes in the first 20 minutes. Their focus has now been to add video as quickly as possible to stories so as not to miss that golden period (it’s automated using a firm called Oovvuu).

“Podcasts have a WTF problem” – with WTF in this case being “where’s the females?” In the UK of course this relates to all the male double acts of the Rest is... series and others, but in Europe they were saying that their pods were predominantly listened to by men, even when they had female presenters. What’s going on?!?