The 4 fundamental challenges confronting digital media

We’re having a week off YouTube. Not because we’re on holiday, but because we decided doing fewer things better is likely to grow us faster. Our next video had quite the production revamp, and it’s also our most ambitious. It’ll be released early next week so you’ll get the link in your inbox shortly after.

While we’re having a fallow week on the Big Picture I wanted to write down some musings on the fundamental challenges facing digital media. Our consultancy works on content workflows, usually at publishers, and it’s become clear that many are pondering an existential threat in the face of the AI wave.

Last week’s video focused on how the Internet enshittified itself with a scaled ad model and network effects, which essentially displaced media brands. I’m also starting to wonder about the essential relevance of many brands in the new AI powered environment. 

While I haven’t seen their latest financial results, from the outside, premium publishers like The Financial Times and NYT look strong. In short, subscription models work. I subscribe to 3 different premium publishers and feel rather more enlightened than I would scanning the news for free. 

That free model is appearing close to over. Digital media brands, particularly in broad lifestyle categories, are going to see their revenues wane and their relevance go with it. I cannot see it any other way unless they get their readers to pay for their content and focus more on broadcast formats like video and podcasts. Few do, and the clock is ticking. I don’t think the same about print, which on the whole seems to doing be relatively well. Declining, yes, but not a cliff edge. I’m sure we’ll still see a newsstand for many years, partly due to the Milli Vanilli Effect we’ve written about before. But a cliff edge is facing the free to access digital model and I wanted to lay out the reasons why.

We’ve written a three part series on AI’s impact on search so won’t go into too much detail here (available in our subscriber area). Initially the SEO chatter focused on the potential of ChatGPT to replace Google as the web’s apex information retriever - but that predictably died away, for reasons contained in our explainer on the matter. The bigger problem was that Google ‘went AI’ and considerably expanded its rollout of AI Overviews and AI mode in the last 18 months. This week Digiday reported:

This is an astonishing collapse. Publishers are seeing well over 20% year on year declines in search traffic, sometimes reaching above 50%. The CEO of Conde Nast is planning for a zero search traffic world. This is the clearest problem for digital publishers offering free to access content, but I fear few will adapt to the paradigm fast enough.

For many, it requires a complete rethink of the digital product set. Being in an SEO team at a publisher right now would feel like playing a season for a football team knowing half their points will be deducted. It sucks, but Google will not play nice and buckle to publisher pressure. AI Overviews are here to stay, sucking the traffic out of publishers like a vampire. 

AI is moving fast and most people are falling behind. 

The Rundown AI keeps you ahead of the curve. 

It's a free AI newsletter that keeps you up-to-date on the latest AI news, and teaches you how to apply it in just 5 minutes a day.

Plus, complete the quiz after signing up and they’ll recommend the best AI tools, guides, and courses — tailored to your needs.

Sign up to start learning.

The ability for AI to digest, summarise and remix content is a looming problem. Reuters’ Digital News report author Nic Newman warned of Claude’s growing capabilities last week at the Media Voices Publisher Summit. I don’t see this as widespread just yet, as advanced AI features like connectors and skills are used by only a small proportion of total AI users - I’d hazard well under 10%. 

But let’s consider what it could potentially mean. It is entirely plausible that something of Hollywood movie standard could be generated by a single prompt within 2 years, combining a long horizon agent with multiple image and video models. It would take a long time to generate, perhaps over 24 hours, but it’s still frighteningly likely.

This makes remixing text content seem like child’s play. Most large publishers are blocking LLMs from accessing their content, but they can’t dodge everything. If ChatGPT has access to an email inbox, it could remix all the publisher newsletters into an audio summary, removing any promotions or ads. It’s not just the remixing of publisher content that matters here, but the ability of consumers to create their own at extremely low cost, which leads to the next point. 

This is one that is curiously absent from the debate. I go to publishing conferences a few times a year, and other than when I talked about it on a panel, I’ve never heard it mentioned. Social media made web distribution far easier. Hardware innovation drove down the cost of production. AI will bring this closer to zero, where the only cost for Hollywood movies on demand is whatever the token prices are and the few seconds to type in a prompt. As I said, give it two years.

AI in itself does not create a media brand. Faceless AI powered YouTube channels may be having a day in the sun, but it won’t last long. But publishers need to wake up to the fact that the main competition is not their few rivals on the newsstand, but the many millions who are now competing for attention across social media, particularly YouTube, and podcasting. Unfortunately for publishers, many unscrupulous ‘creators’ will also use whatever means necessary to gain attention, from copyright theft to using AI to write anything - often with little reputation to ruin in the first place. They are also much more likely to embrace AI, and quickly, which will likely mean the Creator Economy absorbs considerably more attention in the next decade than the preceding one. 

When I set up Absolutely Agentic we had a search first website proposition. The above changes to the environment meant this was quickly abandoned, and now YouTube counts as the real engine of the media business - driving revenue from ads and referrals into our newsletter and then subscription. YouTube is likely to considerably increase its already large share of TV viewing over the next 5 years. It already leads at around 13%, but what happens when this moves to 30%? This is one prediction I see as very likely, which is why we invest in YouTube over anything else. 

Joining these strands together is a fundamental lack of belief in AI. In many ways, chatbots were released too soon. When ChatGPT was launched in late 2022, it was kind of useful, but also peppered with hallucinations and the writing style was weak. Jump forward four years and the landscape is entirely different - but the attitude towards AI as enabler remains stuck in the early days. 

This comes down to debates about whether AI can write solid outputs - it absolutely can with the right processes in place. I write that in bold because the overwhelming sentiment, from social media laments to publishing houses, is that it can’t. It doesn’t get tone. It uses too many em—dashes. The list goes on. But all of these ‘out of the box issues’ can be filtered out with a solid system prompt. We have our own for subscribers here. 

Text content is already ‘liquid’ and it won’t be long before multimedia goes the same way. Motion graphics, once a hazardously expensive enterprise, can now be created in automated workflows to great effect at a fraction of the cost. 

This brings into view the urgency of AI upskilling, retraining, and refocusing of job descriptions to use LLMs and media models as fundamentally useful tools in content creation. 

I’ve outlined the fundamental challenges, but what can digital media publishers do to react? Providing this in detail amounts to a whole other essay, but can be summarised:

Make everything register to access at the very least, and attempt to get paying customers over flighty search audiences, which will dwindle. Growing newsletter subscribers is absolutely essential.

This will likely require a strategic shift towards journalism that has lived experience at its heart. Commoditised news will be killed by AI, but AI doesn’t live in the real world. Publishers will have to offer a deep perspective and consider why the Creator Economy is growing. In tandem with that, get on YouTube, and invest in it properly. Get a recording studio and make podcast and video recording a straightforward process. It’s simply too difficult at too many publishers as it stands.

But most of all, ask the strategic question. What does the business of publishing look like when anything can be created by a prompt? Reskilling, adapting job roles and fundamentally increasing AI access and ability is critical. 

If you’re at a publisher and want to talk more about the contents of this newsletter, then get in touch - [email protected].