Between AI-assisted content, online reputation and platforms that keep more and more traffic for themselves, new balances are emerging that are set to change the way business is done online
Welcome back to a new episode of SEO Confidential, the series where we give a voice to the world’s most authoritative SEO experts to understand where search marketing is really heading, with no filters and no clichés.
This week’s guest is Saijo George, an Australia-based SEO and GEO specialist with a career in the industry long enough to have seen it reinvent itself several times: best practices that turned into penalties, yesterday’s black hat tricks transformed into official Google features. Today he leads the AI-driven SEO offering at Pathfinder Marketing, working with clients ranging from small businesses to enterprise marketplaces with millions of users.
His strength? He doesn’t trust theories: he tests, measures and bets only on what delivers concrete results.
In this interview, Saijo tackles some of the most pressing questions for anyone doing business online today: the real risk of AI in content, the decline in organic traffic, the future of commerce inside AI platforms, and how to build a brand that survives as Google keeps more and more value for itself.
If you run a website or an e-commerce business, don’t miss his insights, because they affect you directly.
Get your website ready for AI agents: build a distributed presence beyond the boundaries of your domain
Everyone keeps talking about the “penalty risk” linked to AI, even though data shows that over 80% of top-ranking pages in search results already use AI-assisted content: isn’t this fear fueled more by consultants and agencies than by Google itself?
“AI-assisted content” and “low-quality, unsupervised AI-generated content” don’t carry the same risk profile, and Google treats them differently. The 80% statistic is real because AI-assisted content – help with research, drafting, editing support with a human involved in the process – ranks well.
But Google’s penalty for scaled content abuse is also real and has been sent out to publishers. It targets the other end of the spectrum: content generated and published without supervision, in large volumes, that no human would actually want to read. So it’s not so much “consultants making up a risk” as it is “a risk that has been wrongly framed as being about AI use, when it’s really about quality and supervision“.
I don’t think it’s 100% “human” versus “AI” anymore: the real issue is content quality.
When do you think AI will stop being a competitive advantage and start becoming a real problem?
When it comes to creating content – and not just blog posts, but also websites, landing pages, ads, videos and so on – AI has significantly changed many people’s workflows, and that’s largely a good thing.
The flip side is the sheer volume of content that is simply rephrased from existing articles. This large-scale regurgitation is the real problem, and I don’t think it’s a future problem; it’s already happening right now.
In your day-to-day work, what do you use AI for the most: to improve the quality and depth of your content, or mainly to speed up production and scale volumes that would otherwise be hard to manage?
Both, but more on the quality and research side. I use it a lot to summarize large amounts of content into digestible chunks, to surface angles I might have overlooked, and to support the research process in general. The speed and scale benefits are real, but they tend to come as a by-product of using it properly for quality, rather than as the main goal.
People talk about the “decline in organic traffic” as if it were a technical SEO problem, but it looks more like a change in the rules set by Google: isn’t the real issue that Google now keeps more value for itself and sends less traffic to publishers?
Yes, and I think the split is fairly clear-cut depending on the type of content. Purely informational sites and affiliate sites, which act as middlemen between the end user and the product, will feel the impact. Most of that content has already been absorbed by large language models, so the answers now live in the training data and, as it expands, the clicks going to publishers shrink. E-commerce and services like plumbers and electricians still need someone to actually do the work, so for now they’re safer.
If the game increasingly revolves around brand and reputation, how do you concretely build that authority: proprietary content, digital PR, reviews, editorial presence… what really makes the difference today in measurable terms?
Ultimately, the goal is to become the source that other players in your industry cite, which means original research, digital PR and earning mentions in the publications that large language models (LLMs) process and use for their training. That’s the goal.
In the short term, LLMs also rely heavily on user-generated content (UGC) when fact-checking search results, so an easier way in is to embed yourself in those communities and conversations first, and then expand your brand’s presence across the wider web from there.
The mental model shift is the important part: AI doesn’t rank pages, it recommends brands. That means you don’t have to do everything on your own website; your presence needs to be spread across the wider web.
Faced with concrete risks such as Negative GEO and reputation attacks, what should a company do in practical terms to defend itself: invest in digital PR, monitor third-party sources, build trust signals… which levers are the most effective in day-to-day operations?
All of that, but on top of it, you need to keep an eye on how your brand is represented across the wider web and within specific models. That means regularly checking what the major large language models (LLMs) say about your brand, which sources they draw from, and which way the narrative is heading.
If something inaccurate or harmful keeps being repeated on UGC platforms, review sites or third-party content, it will eventually shape the models’ outputs. Spotting it early and addressing it at the source is the most practical lever we have right now.
Google wants users to buy directly inside Gemini, while OpenAI aims to influence decisions before the purchase: in the end, who will control the customer, brands or platforms?
I think everyone wants to do all of it. OpenAI recently scaled back its Instant Checkout because user adoption was low, and retailers like Walmart found that in-chat purchases converted at roughly a third of the rate of their own website.
In my view, Google will win in the end. Their UCP (Universal Checkout Protocol, a protocol for integrated purchases and transactions between AI agents and platforms, Ed.) is gaining real momentum; Amazon, Meta, Microsoft, Salesforce and Stripe have just joined the UCP Tech Council, while OpenAI’s ACP is losing steam (Agent Communication Protocol, a standard proposed by OpenAI for communication between AI agents and external services, Ed.).
If people can buy without ever visiting your website, what should a company do today to keep selling: be present inside these platforms, optimize product data and feeds, or build visibility elsewhere? Where should they really start?
They should start with feed optimization: clean, structured, accurate product data is the foundation for everything that comes next, because that’s what AI agents draw on when they recommend or transact.
From there, get your site ready for UCP and WebMCP (Web Model Context Protocol, a protocol that lets AI agents interact directly with websites and web services, Ed.) so you can actively participate when the agents come knocking and, in parallel with all of this, invest in growing your brand’s presence across the wider web. Agents recommend brands, so the more you are cited, reviewed and mentioned across the ecosystem, the more likely you are to surface in the first place.
With large language models (LLMs) entering the commerce space, isn’t there a growing risk of losing trust in the answers? If recommendations are increasingly driven by commercial logic, how will users be able to tell genuine advice from paid or biased content?
I believe Google and OpenAI fear the loss of user trust more than users themselves do; for them it’s an existential threat, so they will fight tooth and nail to course-correct whenever things start to go wrong. The interesting thing is that the way people consume information has already started to change.
As SEOs, we need to watch how this shift plays out and where we can actually exert our influence. For a long time, it was the ten blue links on Google. Now there are real alternatives challenging that and, honestly, like everyone else in the industry, I’m figuring it out as we go.
No longer “pages to rank” but “brands to recommend”
What emerges from the interview with Saijo George is a clear, if uncomfortable, picture: search marketing is not in crisis, it is simply moving onto new ground, and those who fail to understand this in time risk optimizing for a world that no longer exists.
The practical message for you, if you run a business, is simple: stop thinking in terms of “pages to rank” and start thinking in terms of “brands to recommend”.
When AI tools like ChatGPT or Gemini answer a question or recommend a product, they don’t go looking for whoever has the most technically perfect website. They fish for the brands people talk about, the ones cited by other sites, reviewed, mentioned in communities. If your brand doesn’t exist in the online conversation, it doesn’t exist in AI answers either.
Three concrete things to do right now: keep your product or service data clean and up to date (watch out for inconsistencies and errors!), join the online communities where your industry gathers, and work to earn mentions and citations from authoritative sources.
AI is not the problem. The problem is using it badly or, worse, failing to understand how it is redrawing the rules of the game.
Thanks to Saijo George for the clarity and pragmatism with which he tackled every question.
See you in the next episode of SEO Confidential,
#avantitutta
