Between AI Overviews, zero-click results, and an increasingly unpredictable Discover, Trisolute’s News SEO Strategist explains how the relationship between journalism, platforms, and readers is changing
Welcome back to SEO Confidential, the column where we talk with the world’s leading experts in SEO, GEO, and search marketing to understand where search is headed and what it really means for anyone publishing content online.
Today’s guest is Joy Johnston, one of the most authoritative voices in News SEO worldwide. Joy is the News SEO Strategist at Trisolute News Dashboard, a go-to tool for measuring and optimizing the visibility of news publications on Google.
She brings over 25 years of SEO experience and more than 12 years spent directly in newsrooms, where she helped shape search strategies, content distribution, and audience growth. She was a co-keynote speaker at News Reach Con 2025 and led sessions dedicated to news SEO at ONA23. Every week she publishes her LinkedIn newsletter Don’t Panic: It’s Only Data, which you can find directly on her profile.
With Joy, we talked about a topic that concerns anyone thinking about their online visibility today: what happens to publishers in a world where Google is becoming an answer engine more than a search engine, how the relationship with readers is changing in the era of AI Overviews, and above all, which concrete strategies can still make a difference.
An interview packed with useful insights, well worth reading all the way through.
But enough with the preamble, here’s what Joy told us!
From the organic traffic crisis to the risk of “AI slop,” journalism tries to redefine its value
For years, Google traffic was the growth engine for many online publications. Today, with AI Overviews and direct answers, do you think the click-volume model is becoming structurally unsustainable for many publishers?
Yes, the traditional search model publishers have relied on is changing rapidly because of AI. With Google shifting from a search engine to an answer engine, publishers now face a search landscape where their content gets synthesized by bots and clicks are dropping sharply. We haven’t quite reached “Google Zero” yet, but Google I/O 2026 made it clear that AI search is now Google search.
This has created enormous uncertainty and anxiety inside newsrooms. The decline in search visibility comes on the heels of the collapse in social media engagement, which happened as social platforms progressively walked away from the relationships they had built with news publishers. With the two main content distribution channels drying up fast, where can publishers turn now?
Publishers like Condé Nast and People Inc. are gradually moving away from search as a reliable content distribution platform, choosing instead to invest in subscription growth, licensing deals, and solutions outside the traditional platforms.
There’s a lot of noise out there, shiny new tools, and heavily AI-driven companies launching products that may or may not turn out to be truly sustainable solutions.
More and more people read the answer from ChatGPT or AI Overviews without ever opening the original article. From your perspective, is journalism really at risk of losing its direct relationship with readers, or can newsrooms still carve out a new role in this ecosystem? And if so, what should they focus on?
I’ve been thinking a lot about what Nikita Roy said during her talk at ONA25: “The article is no longer the fundamental unit of journalism in an AI-mediated world.”
As painful as it may be for some journalists to accept, I agree with Roy that journalism’s traditional formats are being disrupted and rethought. Articles will continue to exist, but their distribution in full is no longer guaranteed. Online content had already been turned into snippets and summaries long before AI Overviews and AI Mode came along. And over the past decade, video and audio content has grown significantly. I’ve been in the news industry long enough to have seen content formats change many times: blogs, newsletters, video, short-form video, and so on.
Every change has met resistance. Early in my career, a publishing company I worked for announced that the homepage would be converted to a blog format, when it had previously been structured as a page of articles. Some writers quit because they considered blogs unprofessional. I wonder what became of those people, considering that blogs went on to dominate the Internet in the years that followed. Today I’m not even sure readers really distinguish between an article, a blog post, or something published on Substack.
AI-powered search highlights a fundamental difference in how an article from a news outlet is treated compared to, say, a pair of shoes from a fashion retailer. A news article can be broken apart by Google’s AI Overviews and AI Mode to pull out specific quotes and passages, while a pair of shoes will still be presented as a complete product.
AI-powered SERPs are sparking a debate in newsrooms about which traditional search KPIs still really matter and which new indicators need to be defined. Even before AI features started dominating Google’s SERP, there was already a conversation in the SEO community about the value of impressions versus clicks. I’ve always believed both are essential for news publishers. Impressions signal brand authority and trust, but the purpose of a news article is to be read. Anything short of that is a failure.
What journalism risks losing today is the visibility that Google’s traditional SERP delivered in a relatively stable way. At the same time, this could push publishers to find new ways to build direct relationships with readers: newsletters, subscriptions, apps, live events, and more.
Publishers should take a hard look at their brand, their target audience, and their editorial mission to figure out which content to focus on going forward. SEO and audience development professionals can help publishers recalibrate their strategy, aligning content with the brand’s DNA and making sure it comes through clearly in every product they offer.
Many publishers are starting to use AI for headlines, summaries, translations, SEO, and even full articles. In your view, at what point does automation genuinely risk cheapening the value of journalism?
There are smart ways to bring AI into the repetitive, tedious tasks in newsrooms, with relatively low risk. We’ve already seen concrete examples, like using AI to transcribe board meetings or generate recaps of local events. This kind of implementation can be especially useful for small newsrooms operating with limited staff.
At a more advanced level, we’ve seen publishers create AI-generated article summaries and use AI to personalize news feeds. In my view, these uses of AI are reasonable and represent the kind of experimentation more publishers should explore, as long as human oversight and transparency about its use are guaranteed.
Where AI does diminish the value of journalism is when it’s used to generate full articles with little or no oversight, producing what we might call “AI slop,” meaning shoddy, carelessly generated content. We’ve already seen what happens when staff become too dependent on AI without careful fact-checking, as in the case of the Chicago Sun-Times’ “hallucinated” book review section. Mistakes like these risk further eroding readers’ trust in a news outlet. And even if AI managed to keep errors to a minimum, it would still lack the authentic human voice and personal perspective that are essential elements of journalism.
To give an example, a newsroom with limited staff might decide to automate sports score recaps. What cannot be replaced, though, is the experienced sports journalist who can offer original perspectives and insights that fans look forward to reading after every game. Clearly distinguishing between what can be automated and what requires human added value will be critical in the years ahead.
There is a lot of talk about a crisis of trust in the media. Do you think AI can help journalism regain credibility, for example by freeing up time for investigative work and fact-checking, or does it risk making the problem even worse?
It’s hard for me to be optimistic when it comes to rebuilding trust in the media. That’s driven not only by AI but also by society’s growing polarization and its increasingly rigid attachment to echo chambers, closed information ecosystems where people tend to engage only with opinions that mirror their own.
Fact-checking is a topic I care deeply about: in my previous newsroom role I developed best practices and organized training on this front. Unfortunately, fact-checking efforts have proven largely ineffective at improving trust in the media; in some cases, they have even produced the opposite effect. The consequences reach well beyond the news industry and are affecting governments and institutions around the world. It’s easy to blame the algorithms, but at some point individuals also have to take responsibility for the content they choose to consume and believe.
That doesn’t mean publishers should wave the white flag, though. Without a shared consensus on what constitutes a fact, a healthy society cannot function. AI tools could help make fact-checking more efficient and present the results of those checks in a form that’s more effective and easier for the target audience to understand. Newsrooms could use AI to identify the topics that generate the most negative sentiment among readers and adopt targeted strategies to strengthen trust when covering those subjects.
AI personalization features, though, like Google’s Preferred Sources, risk pushing people even deeper into their information bubbles, relying exclusively on publishers that confirm their worldview.
More and more publications are experimenting with AI-generated or AI-assisted content. In your view, should a publisher always explicitly disclose when a piece of content was created with the support of AI? Where should the right balance sit between transparency and the risk of breeding distrust in the reader?
Surveys suggest readers want to be informed when content was generated with the help of AI, and the real balance lies in how that transparency is handled. There are discreet approaches, like watermarking, that make it possible to signal in a non-intrusive way that an illustration used in a news article was AI-generated. On my personal blog, for example, I note which model I use to create AI-generated images. I don’t think it’s necessary, though, to publish in full the dozens of lines of prompt fed into the LLM to get that illustration.
When it comes to articles, it’s important to disclose meaningful AI involvement in the final product, for example if it’s used to generate a bullet-point summary or a data table. In my view, though, there’s no need to specify that a journalist used an LLM to organize their working notes. Every publisher will need to define its own AI policy, and it should be made public on the publication’s website.
The bigger problem arises when third parties implement AI technologies and the newsroom loses control over how its content is presented. Google’s AI Overviews and AI Mode can summarize an article incorrectly or pull a passage out of context.
Google is also experimenting with using AI to rewrite headlines within Discover. I’ve flagged several cases where Google’s AI system changed the headline of an article about an upcoming sports game, even displaying a final score when the event was still hours away. The disclosure that it was AI-generated content was only visible if the user interacted with the Discover card, while the publisher’s name was prominently displayed, pinning the error on the publisher rather than on Google. And yet there’s little publishers can do in these cases other than report the problem to Google.
Today many publishers are pursuing individual deals with AI platforms and big tech players to protect or monetize their content. But news isn’t a product: it’s a common good that shapes elections, public health, and people’s economic choices. Do you think a more collective negotiation is needed, perhaps backed by institutions? Should Google and the AI platforms be required to sit down at the table with publishers?
Large publishers will continue to lead negotiations with Google and the large language models (LLMs). For nonprofits and small publishing companies it will be much harder to get a seat at the negotiating table, let alone strike a licensing deal.
I have nothing against publishers seeking deals with Google and the LLMs to monetize the content those platforms use to power their services. What worries me, though, are the consequences of these deals for search algorithms. It will no longer be about which publisher offers the best content, but which one has the most favorable business deal with Google or a given LLM. That could make search engine optimization efforts less relevant and risk further reducing the visibility of small news publishers.
AI systems are training on content created by publishers, journalists, and creators, often scraping information without paying them anything for it. So let me ask: if they lose traffic and revenue because of AI-generated answers, what incentive will they have to keep producing original, up-to-date content? Isn’t there a risk of a vicious cycle where data quality deteriorates and, as a result, the AIs themselves get worse too?
This is something I have experienced firsthand: a book I published was used by Anthropic without my permission.
Yes, there is a growing threat that AI could eventually run out of new content to learn from. The worst-case scenario is one where AI tools start feeding mostly on content created by other AIs, creating a vicious cycle of “AI slop,” meaning low-quality, artificially generated material, that would be very difficult to fix.
The most immediate effect, though, is a different one: a growing number of publishers will start locking down their content, putting up paywalls or shifting to subscription-based models. We know AI systems are notorious for often finding ways around these strategies and scraping content anyway. If publishers actually managed to block AI bots, those bots would likely end up feeding more heavily on low-quality content from less reliable sources. Another consequence could be that the most authoritative journalism becomes less accessible to the general public.
It will be interesting to watch the impact of the provisions introduced by the UK’s Competition and Markets Authority, which will allow British publishers to opt out of Google’s AI search products and prevent their content from being used to train AI models. How many publishers will decide to opt out? Will the quality and variety of the answers offered by AI Overviews and AI Mode suffer? Publishers around the world will be watching very closely.
Many publishers are becoming increasingly dependent on Google Discover: isn’t there a risk that building a business on an unpredictable feed ultimately means renting your audience instead of owning it?
Absolutely. Google Discover has become something of a consolation prize for many publishers as they watch their SERP traffic progressively dry up. However, Google Discover has always been an extremely volatile and unpredictable platform, one that should never become a publisher’s main source of search traffic. And yet that’s exactly what has happened in recent years: some publishers have seen between 80% and 90% of their organic traffic come from Discover.
It’s a potentially disastrous situation.
Google Discover can certainly drive significant traffic volumes. But if publishers have learned anything over the past decade, it should be this: relying exclusively on a third-party platform to reach your audience is risky. Whether it’s Facebook abruptly shutting down its dedicated news product or Google filling the SERP with zero-click AI features, publishers need to reduce their dependence on these platforms for content distribution.
Google Discover seems to increasingly reward content that captures the immediate click: isn’t there a risk that publishers end up sacrificing quality and editorial depth to chase whatever works in the feed?
Clickbait has always been a problem on Discover, but what I’ve noticed recently is a growing presence of hard news in the Discover feed. If this trend takes hold, it could bring both upsides and downsides. On one hand, major global news publishers could pick up additional Discover traffic from content that’s already at the core of their SERP strategy. The ideal is when a piece of content manages to perform well across all of Google’s surfaces. On the other hand, niche and lifestyle publishers may be forced to rethink their Discover strategy if they see performance drop because of this shift toward hard news.
Another problem is that Google is using Discover as a sort of testing lab for AI. I’ve said before that AI-generated headlines are terrible and automatically produced summaries can be misleading. Discover is essentially a hybrid between search and social media, so headlines, images, and summaries need to be designed to earn that precious click. By altering how news is presented in the Discover feed, AI is making it harder for publishers to keep relying on the tactics that used to work well.
Ultimately, whether we’re talking about AI search features, the SERP, or Discover, the central point remains the same: publishers need to focus on their audience and on content distribution models they can control directly, instead of chasing such an unpredictable algorithm.
Traffic isn’t rented, it’s built. And, today more than ever, on foundations you own
Our conversation with Joy Johnston was a rich one, at times contrarian, and extremely useful for anyone working with online content. Whether you’re a publisher or a professional seeking visibility in an ecosystem that changes every six months, you simply cannot afford to ignore what Joy told us.
The model that worked for a decade, built on organic traffic, predictable SERPs, and distribution delegated to the platforms, is not coming back. This is not a temporary crisis, it’s a structural transformation. And anyone still building their editorial or content strategy by betting everything on Google, whether that’s traditional search or Discover, is essentially renting their audience instead of owning it.
The direction to take is clear: invest in a direct relationship with readers, be selective in how you use AI, demand transparency, and have the courage to redefine your KPIs instead of chasing metrics that no longer mean what they used to in the new ecosystem.
It’s not an easy road. But it’s the only one that leads anywhere.
A huge thank you to Joy Johnston for her time and for the generosity with which she shared her experience: if you’re not following her on LinkedIn yet and haven’t subscribed to her newsletter, I recommend fixing that right away.
We’ll be back next week with a new international guest and new questions. See you soon.
#avantitutta
