AI search will (likely) make things worse for all but ultra-premium media

Over the medium term, this will likely impact our choice of media as both advertisers and consumers, further degrading the already-challenged situation for all but the most premium titles.

Search is practically all news publishers’ most important single source of traffic. It now seems inevitable that AI chatbots will replace search as our means of retrieving certain types of information. My own Google v ChatGPT history suggests I’m already using it instead of searching where I think it will be quicker (eg, ‘Can I block iPhone apps at specific times of the day?’) and queries where the combination of different sources is beneficial (eg, ‘Give me some ideas for a baby gift’).

Search queries are usually divided into three categories - navigational (searching for a particular website or brand), informational (looking for info on a topic) and transactional (buying a product/service). Because navigational searchers would find your site anyway and transactional searches make up a small proportion of overall traffic, the informational queries matter for news publishers.

The data tells us that the kinds of informational queries driving traffic are different for different categories of publishers.

For The Sun, the top ten informational searches over the last 28 days include ‘Arsenal,’ ‘Luke Littler girlfriend’ and ‘Love Island.’

By contrast, The Times - its News UK stablemate - includes ‘Ukraine,’ ‘best savings account,’ and ‘Farfetch,’ the struggling online luxury retailer whose share price has plummeted in recent months. Putting aside media snobbery and football allegiances, there are important differences between these types of queries that will, I believe, be a leading indicator of publisher resilience to AI disruption.

For the queries driving traffic to The Times, the user is more likely to use the information in a higher-stakes onward setting. Most obviously, it involves selecting between different savings accounts. For these queries, the source matters. Investment decisions are not going to be made based on information from an unknown website. An AI chatbot just isn’t that useful in this context. If the user was given a single, natural-language response rather than a list of links for many users, it would provide less utility than conventional search results.

That is unlikely to be the case for a lower-stakes search term. A chatbot response to a query about Arsenal, summarizing the latest team news and displaying results and fixtures, could well satisfy the user’s informational need without visiting a destination website. While this isn’t a binary point - and will vary for the user, their affinity to particular news brands and the query - it’s beyond doubt that the source matters less for this type of query.

Google’s search product already reflects these different user needs. The amount of information it presents on the results page changes dramatically depending on the search term. For Arsenal, we are served lots of Google-curated content. For the ‘Taiwan election,’ we see almost none.

And that’s also because Google doesn’t want responsibility for those high-stakes queries. Particularly if there’s a political dimension or controversy, it’s likely to ensure guardrails remain rigidly in place around those topics, at least until legal proceedings testing AI developers’ responsibility for outputs are settled.

So, the big picture here is that up-market titles will likely be somewhat insulated from the traffic-eroding effects of AI integration into search, whereas mid-market and down-market titles will bear the brunt. And this is doubly worrying because they are going into this in a less resilient place. Firstly, they rely more on search for traffic; the average UK tabloid relies on organic search for over 50% of traffic. That figure is around 30% for broadsheets. Secondly, their models are dominated by advertising revenue, which correlates to traffic.

Over the long run, we could see a further hollowing-out of the market for news.

At the top end, a small number of high-end titles will serve paying premium audiences. The near-zero marginal cost of sale and economies of scale are already acting as forces to further concentrate this segment. Meanwhile, mid-market and down-market outlets will struggle to sustain their newsrooms even more than they do now. That gap could be filled by misinformation, disinformation, or just low-quality content.

David is an independent media and marketing consultant and director at the FT. These words are his and do not reflect the views of his employer.