Barry Adams sees Google Discover as the most tempting shortcut, but warns: it is unstable and prone to volatility. The real long-term path is still diversification
SEO is going through an unprecedented mutation: AI Overviews and AI Mode are changing the way we compete on search engines, while Google makes it increasingly difficult to access the data that really matters.
For those building their business online, it is no longer about “staying up to date”, but about understanding – and doing so quickly – how to move across such precarious, unstable ground.
This is the need behind SEO Confidential: a series of interviews with the brightest minds in international SEO, designed to offer solid analysis, strategic perspectives and useful ideas to seize the opportunities of the new online search.
Today we are delighted to talk with Barry Adams, one of the most experienced SEO consultants for publishers worldwide. Barry has over 25 years of experience, founded Polemic Digital, runs the site SEO for Google News and has worked with names such as The New York Times, The Guardian, Euronews and News UK.
With him we discussed the growing opacity of Google’s data, the risk of AI draining vital traffic away from publishers, the need to redesign the user experience, and the strategies for keeping a brand memorable even when content gets synthesized by AI.
How can you stay front and center when Google shrinks the space available? And how much is it worth talking about AI optimization if the new “answer engines” tend to paraphrase rather than cite?
These are the questions Barry answered without mincing words, in a conversation aimed at understanding which strategies can make the difference today.
“The numbers have always been guesses, the data has always been unreliable”, Barry Adams told us
Today Google offers less and less transparency on the data that truly matters to SEO professionals. If Search Console makes it impossible to distinguish traffic coming from AI Overviews from traffic generated by AI Mode, how can informed decisions be made? And, above all, what is the point of talking about SEO strategy if the metrics to base it on are missing?
The thing is, SEOs and digital marketers have been a bit too credulous with the data they analyze for far too long.
In reality, every data source is imperfect: the numbers are always estimates, never absolute truths.
If you truly believe that Google Analytics 4 reports faithfully represent every single visitor and accurately reflect users’ real behavior on your site… well, then I have some fantastic beachfront property in Austria to sell you!
Unreliable data is nothing new: it has always been part of the game. And yet, this has never stopped professionals from doing their job.
I remember when Google removed organic search keyword data from GA (the so-called “not provided” problem) and SEOs thought it would kill our industry. Guess what? Nothing happened. We simply adapted and learned to deal with it.
Another example: most news publishers rely heavily on having their articles shown in the Top Stories carousel at the top of Google’s search page for news-related queries.
However, Google has never provided data on Top Stories visibility. Ranking in that area is treated exactly like ranking in general organic results, with no dedicated report or filter.
That has never stopped publishers, though: they kept optimizing their sites for Google and using third-party tools to monitor Top Stories. It will be the same with AI Overviews. We will find ways to track them and adapt our SEO strategies accordingly.
At some point, a separate report for AI Mode will probably arrive too. Right now, AI Mode is a distinct tab and, historically, Google tends to provide specific filters and reports in Search Console for each separate tab in the search results.
To what extent is AI Mode a leap forward compared to AI Overviews, both in technological terms and in its potential impact on traffic?
AI Mode seems to be built on more advanced technology than AI Overviews. While the latter generate answers in real time by drawing mostly on Gemini’s training data, AI Mode uses advanced query fan-out techniques to look for up-to-date sources and then synthesize them.
This allows it to answer questions about current events too, something AI Overviews do not do, because Gemini is not yet trained on the latest news. And this is exactly where the problem lies for publishers: with AI Mode, users can get the latest headlines and then ask follow-up questions, receiving ready-made summaries without ever opening an article. In practice, AI Mode is capable of summarizing the news for you.
AI Overviews do not appear for queries that show a Top Stories box, which is why news publishers have not been significantly affected so far. With AI Mode, however, the scenario could change radically, with potentially heavy traffic losses for all publishers, not just news publishers.
That said, Google probably does not want to cut off the flow of visitors to publishers entirely and is looking for a balance: integrating AI into search without reducing traffic too much. The most recent experiment, Web Guide, goes in this direction and seems to signal a desire to find a middle ground between AI Mode and traditional search.
Barry, you have pointed out how users’ frustration with the experience of modern websites fuels the use of AI Mode. Do publishers still have room to improve usability, or are we already too far into this dynamic of “disintermediation”?
Publishers always have room for improvement when it comes to user experience. If a publisher wants to build audience loyalty, an excellent UX must be one of the pillars of their strategy. Too often a website’s commercial needs have taken precedence over its users’ needs, and that dynamic is unsustainable. That is why AI Mode is so appealing to users, as it lets them interact with the web’s content without the obstacles the average website puts in front of its users.
If publishers really want to build a loyal audience, they will have to create experiences free of friction and slowdowns for their readers. Paywalls come to mind (which can remove the need for advertising), as do dedicated apps. Chasing ad-monetized clicks is a business model with ever-diminishing returns and, in my view, it should largely be abandoned.
In AI Mode even the news gets summarized without direct citations of journalistic sources. Do you think this marks the beginning of a definitive marginalization of journalism in Google’s results?
AI Mode does provide links to sources, but users are unlikely to click on them as often as they do with traditional search results. I do not think Google considers journalism a marginal activity; it is more likely that programs will be launched to compensate publishers when their content is used in AI Mode.
Licensing models could emerge, perhaps as an evolution of the current Extended News Previews program. Google wants to support journalism, but not to the extent publishers would like.
Among the survival strategies you suggest – Discover, diversification and direct branding – which do you think is realistically the most viable for the average publisher? And which is the most sustainable in the long run?
The most tempting strategy is Discover. It is relatively easy to optimize for and can deliver immediate results. However, Discover is also one of the least solid channels, with a high degree of volatility, great vulnerability to algorithm updates, and it is often prone to rewarding cheap clickbait.
I advise my clients to optimize for Discover only insofar as it does not compromise the quality of their journalism in any way. But that is easier said than done.
I think diversification is the most sustainable strategy, but also the hardest one. It means investing in different channels, different content formats, different modes of consumption, and thinking about your editorial output in a fundamentally different way. I suspect most publishers are unwilling or unable to do that. The ones who are will survive and thrive in the AI era.
Diversification comes up often, but for many small and medium publishers getting into Discover or Top Stories is increasingly difficult. Has Google already decided who gets to stay visible and who does not? Is the market now closed, and will it become ever more so?
The market is not closed, but it is a zero-sum game: a new publisher can find space, but only at the cost of displacing someone else. Doing so is hard and requires continuous effort.
Competition is fierce and good intentions are not enough. Standing out in the limited space of Top Stories and Discover requires constant adaptation, a focus on your strengths and on the characteristics that make your editorial brand unique, plus the ability to put your audience’s needs ahead of the publisher’s own.
Many publishers chase AI optimization with technical or semantic approaches. But in your view, is it really possible to optimize for a system that, as you put it, “regurgitates content more than it cites it”?
I would question the validity of these “AI optimization” tactics. Almost all of this advice is based on an incomplete understanding of how large language models (LLMs) work, or it is recycled SEO advice with a different label (GEO, for example) so the agency can charge more for the same service.
I think optimizing for AI is much like optimizing for Google. You can try to keep up with the algorithms by hunting for “cheat codes” and tricks to gain traffic, or you can try to win over your audience and let intermediaries like Google or ChatGPT decide whether to give your content visibility. The first approach has never led to lasting success. The second has.
If publishers shut down, original content disappears. But if content disappears, AI is also left without data to draw from. Is this not an unsustainable model by definition? Is Google destroying the well it drinks from?
Let me play devil’s advocate: Google does not need a billion websites to crawl, index and rank. Every possible piece of information and every possible service or product is already available hundreds of times across different competing websites. If the web were ten times smaller, Google would still have everything it needs to provide search results and AI answers to its users. A smaller web would suit Google just fine, reducing the number of URLs it has to crawl and index and significantly lowering the costs of its search systems.
But a smaller web would have significant economic and social consequences. And we cannot allow Silicon Valley companies to decide, on their own, how and how much to shrink it. A platform with Google’s power should be regulated, to prevent it, in its constant race for growth, from causing large-scale economic damage.
You wrote that “the worst thing that can happen to a publisher is to be forgettable”. What makes a brand truly memorable today, in the eyes of both users and algorithms?
I think it is different for every brand. A publisher should not make internal assumptions about what its audience thinks of it. The number one piece of advice I would give here is “talk to your audience“. Ask them what they think of you. Ask them why they read you. What they like about you, what they do not like, what they would want more of, what they would want less of. It can be an extremely enlightening exercise and it should be done regularly.
At the same time, your audience does not know everything. Do not be afraid to experiment outside the frame of your readers’ answers. Sometimes the audience does not know what it wants until you show it to them. So, take your audience’s feedback into account, but also experiment beyond that feedback with new types of content and formats.
At the moment AI Mode adoption is low, but you speculate that Google could integrate it more visibly into the SERP or make it the default instead of classic results. If that happened, what do you think the concrete, immediate effects would be for publishers?
If AI Mode became the default search experience, most publishers would suffer an immediate and disastrous loss of traffic. However, Google is unlikely to take such a radical step: it knows full well what the impact would be, and it knows the backlash would be enormous. AI Overviews have already generated a strong media backlash, and AI Mode would provoke an even fiercer one. Google will look for a balance instead.
The company’s real short-term concern is not losing users to OpenAI, but managing to offer an AI experience comparable to – or better than – ChatGPT’s without substantially compromising its own search engine. Anything but a simple challenge.
Make yourself memorable even when AI reduces your work to a few lines
A clear message emerges from Barry Adams’ words: SEO has never been an exact science, and today it is even less so. The illusion of being able to base every decision on perfect data has faded; what remains is the ability to read the signals, adapt, and move with clarity in a market where the rules keep changing.
AI Overviews and AI Mode, as I have told you time and time again, are redrawing the dynamics of search. It is no longer just about holding your ground in the traditional SERP, but about getting recognized and cited by AI systems that synthesize, select and filter information before the user ever reaches your site.
And here allow me one observation: at this stage, optimizing for a search engine no longer means chasing trending keywords, but building an ecosystem of content and brand signals so strong that it teaches the AI to consider you a trustworthy source.
This requires thematic consistency, depth of analysis and a distribution capable of reaching users wherever they get their information. “Post-AI” SEO will be won by those who can merge editorial authority and presence strategy, making sure that, when a question comes up, the algorithm immediately thinks of you as the answer.
That is why the real goal is not “getting clicks”, but becoming an authoritative reference. You need to diversify channels and formats, invest in a user experience free of obstacles that slow down or disturb navigation, and build a brand that sticks even when your content gets compressed into a few AI-generated lines.
The question is no longer “How do I optimize for Google?”, but “How do I stay relevant when Google decides what to show and what not to?”.
Those who can provide valuable answers – and do so consistently – will be the ones artificial intelligence keeps citing and taking into account. Everyone else risks slipping to the margins, even if technically “ranked”.
My heartfelt thanks to Barry Adams for sharing such a lucid, direct vision of what lies ahead. See you in the next episode of SEO Confidential: the journey into the future of search certainly does not stop here.
#fullsteamahead
