From dotcom to AI: has the news industry learned anything?

The latest Ipsos iris data makes for uncomfortable reading for anyone invested in the future of local journalism.

Across the UK’s 75 largest local news websites, average monthly audience fell by 19% year-on-year in April. Time spent dropped by more than 16%. Only a fraction of titles saw any growth in either reach or engagement. Drill a little deeper and the picture becomes even starker. Some brands experienced audience declines of more than 50%, with a handful falling by 70% or more.

Behind those numbers is an unfortunately familiar story - one that has defined news publishing for two decades or more.

History of missing the shift

The internet was originally laughed off as a fad. Indeed, an infamous Daily Mail headline from 5th December 2000 read: "Internet may be just a passing fad as millions give up on it". That disregard was emblematic of wider industry indifference to what in just a few short years would change the world.

By 2008, when I was a reporter working for Central Independent Newspapers, part of the Daily Mail Group, the industry was already undergoing consolidation, cost-cutting and restructuring. I was based 40 minutes away from my patch, in a single office that housed reporters for five or six local rags covering a huge amount of the West Midlands and Staffordshire.

As soon as I arrived, pagination was dropping rapidly, then my paper turned into 60 second news, and finally redundancies arrived and the office got consolidated further, merging with another in Stoke-on-Trent. The global financial crisis can be blamed for a lot of that at the time, of course, but the nails were already in the coffin, they just needed a hammer.

Over the following two decades, newsrooms have continued to shrink, titles have closed, physical offices have disappeared from many communities. All of this has had a cumulative effect on audience relationships. Local journalism has always depended on trust, proximity and relevance. Those things are harder to maintain when resources are stretched and presence is diluted.

Continually playing catch-up

You would think in 2008 we had a website on which to publish our news - alongside a newspaper. We did not. Digital publishing was still in its infancy, even after platforms like Google and Facebook had started to change the distribution model. With the internet, people didn't need to buy a newspaper, they could get their news online.

News publishers were slow to this fact and have been playing catch up ever since. When regional newspaper websites did start arriving, many adopted an advertising heavy approach to try to generate revenue that once largely came from physical newspaper sales - to the detriment of the user experience.

For the last decade or so, local publishers have operated within a simple, if uneasy, trade-off with these social and search platforms. With their giant userbases, the likes of Google and Facebook delivered scale, and while that helped publishers get footfall to their difficult to use websites, they ceded a degree of control over how their content was distributed.

That model was always fragile. It relied on algorithms remaining broadly favourable, or at least predictable. The Ipsos data illustrates what happens when that assumption no longer holds. Titles that had benefitted significantly from sources such as Google Discover have seen traffic fall sharply as those systems evolve.

The rise of the “no-click” audience

Now AI is changing the rules of the game once again. AI-driven search is accelerating a shift towards “no-click” consumption. Users increasingly receive answers directly within search environments — summarised, synthesised, and presented without the need to visit the originating website. From a user perspective, this is efficient. From a publisher perspective, it is a bit of a disaster for their revenue model.

The value of journalism has not diminished, of course. If anything, it has increased in an environment flooded with misinformation and noise. But the route through which that value is realised - the click, the page view, the session - is being eroded. This creates a growing disconnect between the organisations that invest in producing original reporting and the platforms that intermediate its consumption.

It seems the industry has learnt somewhat from the dawn of the internet and regulatory intervention is already beginning to take shape. The Competition and Markets Authority's (CMA) move to require greater transparency and control over how publishers’ content is used in AI-generated results is significant - a recognition that search is no longer just a referral mechanism.

However, the opt-out model the CMA has demanded only addresses the symptoms of the issue rather than resolving the economics behind it. Hopefully, as the watchdog stated: "This will put publishers, like news organisations, in a stronger position to negotiate content deals with Google."

The case for licensing

That is where licensing deals can help - and there are examples of these already taking shape, such as the Guardian Media Group announcing a strategic partnership with OpenAI in 2025. Search and social platforms are now a significant part of the reality of how people access information. AI systems that use publishers' content to answer users' questions should be compelled to return equivalent value to the organisations that created it - and that can be achieved through mutual agreements.

Additionally, these agreements need to encompass better measurement frameworks. If AI-generated responses are drawing on journalistic content, then aggregated attribution (however that is defined) should form part of how reach and impact are understood. Using this additional data, we will get a better picture of the impact journalism continues to have.

There is no escaping this is a new phase in digital publishing and news organisations must avoid being as late to the party as they were with the rise of the internet in the 1990s and 2000s. Previously, it was all about traffic, now it is about value. And while platforms and algorithms will forever continue to change, regulation and adaptation must keep pace.