The discussion around AI and journalism is intertwined with SEO, and with AI changes forecast on the horizon, it’s often cast in gloom.
With Google’s partial launch of AI mode in the coming months, it’s difficult to foresee anything but a reduction in publisher traffic from what has been the most stable source for twenty years. Agents may also make fundamental changes to how users interact with the web - but wholesale changes of deeply ingrained behaviour may take years.
I think these potential changes are looked on too pessimistically. I’m on the fence about AI job destruction and creation - considering that they may well happen in equal measure. I’m 100% in agreement that AI will be disruptive, but there’s also significant opportunity for those who surf the coming AI wave. While rising, it hasn’t broken yet.
A question of value
The big question around AI, SEO and journalism is - does the potential reduction in traffic represent a meaningful reduction in value?
For those heavily reliant on clicks and advertising revenue, I would hazard that this is problematic.
For those offering commercial content that informs buying decisions, this is less so.
For subscriptions, less again.
Possibly for the latter group, it will mean very little at all.
In my experience, AI solves specific problems and I use chatbots largely in two specific ways:
Troubleshooting immediate software problems where AI can provide an answer faster than forums.
Answering low value long tail queries, for which I don’t really want to click into a website to find out.
Both of these types of queries and prompts represent very low value searches. I am likely to go to a website, try and validate an answer, then bounce. That is it. You’re talking 1-2 pages a visit and bounces rates of 90%.
Are SEO explainers journalism?
News websites might rank for keywords like these, but we can hardly call that journalism. To use AI for news is asking for trouble. Asking it for insight led product advice is even worse. I assume that only a small percentage of regular AI users really use it for these things. Instead, they will probably use it for general assistance as my two above points cover.
Writing a load of SEO explainers to tell me when the next election is isn’t journalism either. I say this from experience. As Head of SEO and Social for a national newspaper and later covering the 2024 US election, there was always a trade off between SEO explainers and what could feasibly be done while maintaining quality.
A large newspaper website could easily use AI to create a large directory of pages that relate to news stories. In some ways, automated tag and topic pages have done this for a while. But neither is really high value.
This loss of traffic may not matter
If this traffic was to go, would it really matter? I’m sure alarm bells would be ringing throughout any organisation that lost 20% of traffic overnight. Or even if it was anticipated by some and slowly melted away, the Head of SEO would have to constantly explain why their KPIs were going down. But would this equate to the same amount in revenue?
I very much doubt it.
Perhaps, instead, it would be a wake up call. That a large amount of time spent on low value content to get people into websites is reaching its endpoint. Their queries can be solved by AI. End of. Stop worrying about it. Start putting your energy elsewhere.
Where is that elsewhere? I have a plan.
And it doesn’t rely upon making TikTok videos.
The doom loop of multitasking
My hypothesis is that, from roughly 2007, proper journalism has caught a cold. The cold was brought on by a kind of multitasked overwork driven by the requirements of a media company:
The SEO team needs this. The social media team wants this. The video team has asked you to spend two hours travelling to and from location to present this snippet.
Worse still is the utter hoover of time which are the majority of half loved, half broken CMS systems, whereby journalists (if we can still call them that by now) have to wade in, with little multimedia training, to publish their stories. This was previously not a thing they needed to do, never mind the numerous other things they now have to do:
Proofread and sub, format in CMS, place the right tags, find the right images, cut and resize images on Photoshop, fill out the right taxonomy options, press publish, notify relevant teams.
Does any of this blizzard of multitasking have anything to do with journalism? Is it not just a multitask of multimedia editing that will more likely lead to burnout? Not, there is no research or writing before we reach this stage.
Sensible newsrooms have created a divide between reporting and production, but many don’t have the resources to - or at least think they don’t.
From about 2007 onwards, there were way more sources of information, compounded by many journalists having to switch their attention to attend to them.
Enter AI to end the madness
AI can enhance many steps in this workflow:
Research Research story - AI deep research, document summary and chatbot feedback are all amazing. If you combine these tools with vast sources of information, like court or transcripts or academic papers, you can research exponentially faster.
Format in CMS - automated workflows can do this well already. Details and finesse can be added over time.
Ensure distribution - if a social media update is a summarisation of deeper information (an article) then this can be largely automated.
I leave 'add insight' and 'writing' out of this process because they go together and are best human led.
Add insight means interview people for their view. Physically test what you write about. AI cannot do the things that quality journalism does.
The step is massively enhanced by AI because the journalist does not need to multitask nearly as much.
Writing is combining the gathered sources/insight into a narrative. Sure AI can help here, but I still think it is generally best to write from expert insight. I write about 2k words per week and use AI sparingly.
This is generally what Google is looking for in EEAT, so AI enables journalists to double down on it. Pretty simple really.
The switch to multimedia
If you think that the 'add insight' stage - the interview, the product testing - is really the value add, then the huge decrease in other multitasking enables multimedia recording of this step (video, podcasting) which adds to the promotional mix.
Of course, training needs to happen - but with multitasking at least lessened, there should be more time for this.
Social media algorithms now demand video. The added reach promotes the brand and can be monetised.
As mentioned, AI led search, via AIO or ChatGPT, potentially creates a visibility problem for publishers, but conversely readjusting content strategy to a more human insight led approach is absolutely necessary to stand out.
The future of publishing is less about eyeballs, more about informing buying decisions and subscription.
People will pay for high quality information one way or another.
And meanwhile, I don't believe this reduces the value of Audience teams. These will become more about agentic management (having an amplifying effect) on current research workflows.
Social media specialists can fulfil the necessary (and very interesting) role of multimedia producers. The journalist takes role of presenter, while AI takes care of the summarisation.
I can only see the above, executed well, as a positive.
There are good reasons to be optimistic.
