As AI agents begin to make autonomous decisions and interact with content, building online visibility increasingly depends on trust, accessibility and semantic relevance
Search is changing faster than user behavior, and SEO is no longer just optimization: it is the interpretation of the future.
AI agents are starting to make autonomous decisions, accessibility standards are becoming central again, and branding risks being judged by systems that do not look at pages, they analyze them.
The debate about what to do, how to adapt and what future awaits search is more open than ever, fueled by uncertainty, conflicting impressions and strategic choices that no one can postpone any longer.
This is exactly why SEO Confidential was born: to offer a direct conversation with the people who are redefining the industry and bring readers useful, clear-eyed and immediately actionable perspectives.
For this new episode I invited Jes Scholz, international marketing consultant, leading speaker in the Search landscape and strategic advisor to enterprise companies.
With her we tackle some of the most urgent questions every business owner should be asking today: what happens to the funnel when the “customer” is an AI agent?
Why is usability becoming a survival factor?
What is left of SEO when there is no longer a visible SERP?
And above all, how do you build brand authority if the citation is worth more than the click?
A conversation that does not merely describe what is happening in search, but clarifies which aspects really matter to stay relevant while AI rewrites the dynamics of the market.
A precious opportunity to understand where the industry is heading, how the choices of the new “artificial customers” are changing, and which strategies can make the difference in the years ahead.
It is truly an interview not to be missed, so happy reading!
Here is why your brand needs to win over AI agents: the new frontier of trust
If AI agents start making decisions, purchasing and booking on behalf of users, what happens to the conversion funnel? How can a brand consolidate trust and recognition when the operational counterpart is no longer a person, but an algorithmic model?
Even when the customer is an AI, the decision-making process follows patterns close to human ones. AI systems tend to mirror people’s behaviors and reproduce their logic of choice.
A brand’s ability to be recognized is built before the research phase, because share of voice is formed in advance and guides both the user and the model. A brand already present in a person’s mind or in the AI’s “memory” is more likely to be interpreted correctly, connected to the search intent and shown in a relevant position. This early visibility also increases the chances of surfacing in search engine APIs, and therefore of being selected as the preferred destination by AI agents.
Regardless of the audience, human or machine, the key is to build trust and recognition by meaningfully engaging your audience when they are out of market, so that your brand is relevant when they enter the market.
Your article on Search Engine Land shows that 63% of sites are abandoned by agents due to technical or usability problems. Aren’t these agents simply reacting to problems that human users have been putting up with for years? I am talking about slow pages, broken redirects or other frustrating experiences… Is it really the AI’s fault, or is the problem badly designed websites?
In most cases, it is website usability that fails the test. Many of the issues that put agents off would put humans off too. I believe agents are the best thing to happen to UX in a long time.
Humans weigh the time investment required for further research. They accept “good enough” as a trade-off for time. Agents don’t. They will quickly abandon a site and look for another option if the website’s usability is poor.
You have observed that agents almost always pick the first result, confirming that “being first” matters more than ever. But if choices happen via API, with no ads and no visual results, what concrete role is left for SEO? How can you still influence visibility in a system where there is no longer a real SERP?
Rather than a SERP, it is a SERAPI: a different method of consumption, but with similar principles. SEO still plays an important role because, whether it is the visual search results page or a list of API results, it remains an algorithm returning an ordered list. It is still the SEO’s job to influence which result is displayed first.
Agents read pages in text mode, ignoring images, pop-ups and graphics. Doesn’t this completely upend the rules of web design? In a world of artificial users, are we going back to writing for code rather than for people?
I would see it as a return to writing cleaner code, which is ultimately better for people. After all, which human user likes intrusive pop-ups? Who wants the page to load more slowly to allow for pointless animations? Or worse still, to have 3 different tracking scripts? Developers have often followed popular but bloated frameworks for their own interests, not for the good of users.
But there is more. Text-based navigation encourages best practices such as semantic HTML and accessibility standards like WCAG (Web Content Accessibility Guidelines) and ADA (Americans with Disabilities Act). In recent years these standards have not had a strong business case for investing in code quality, but agents provide one.
(Editor’s note: The WCAG (Web Content Accessibility Guidelines) are international guidelines that describe how to make websites and digital content accessible to people with disabilities, defining technical criteria and best practices.
The ADA (Americans with Disabilities Act) is the US law that protects the rights of people with disabilities and requires digital services to guarantee an adequate level of accessibility as well).
You say that “AI is the new customer”. But if agents make decisions based on structured data and technical signals, don’t we risk creativity and branding losing weight compared to pure technical optimization?
Algorithms have been making decisions influenced by structured data and technical signals for decades. Agents take this concept to a new level of complexity. But I do not see this as a threat to creativity or branding.
Agents follow choice patterns that are surprisingly similar to human ones: they do not analyze every alternative on the market, they select a limited set of options and, just like humans, they often settle. The outcome depends on the brand’s digital availability and its ability to be relevant.
The technical side of SEO helps make the brand accessible and correctly interpreted, but the actual selection comes from signals that indicate preference and reliability. Signals such as clicks, conversions, low return rates (via Google Shopping), positive reviews, social chatter, authoritative coverage from external sources. Technical SEO helps earn the ranking, but it is branding that earns engagement and, ultimately, long-term profit.
With tracking limitations and the opacity of data coming from agents, traffic analysis becomes almost impossible. How can a company truly measure the success of its strategies if the classic metrics – sessions, time on site, bounce rate – no longer mean anything?
Share of voice (SOV) is an excellent KPI for SEO. It takes into account the importance of consistently reaching both in-market and out-of-market audiences. Moreover, when well defined, it is a leading indicator of market share movements, tied to company revenue, which is what a CMO cares about (Chief Marketing Officer, the person who leads a company’s marketing strategy, editor’s note).
The problem with SOV, for AI search, is that there is no shared data on platform prompts. You have to make plausible assumptions about user intentions and you will have a sample limited to the size of the prompts you track. But this is the equivalent of a focus group. As long as you ensure unbiased selection and track consistently over time, the data trends have great value.
Is opening sites to AI crawlers always an advantage? Many publishers fear that allowing crawling by ChatGPT, Perplexity or Gemini bots means losing control over their content and giving away value to platforms that bring no direct traffic. To what extent is this exchange fair? And what distinguishes a smart opening strategy from a simple free handover of value to AI models?
Let’s not fool ourselves, we lost control over content to the platforms decades ago. Yes, artificial intelligence takes this to a whole new level of distribution, since it is not just a teaser, but often a disclosure of the idea in its entirety.
However, this has been a working model for years with content syndication. Similarly, most big brands have invested in building communities on TikTok and YouTube, where content is consumed on those platforms. Neither generates significant amounts of direct traffic. But we understand that being present matters for the brand. AI surfaces are the same.
That said, if you have truly unique content. For example, a large amount of customer reviews. I would not recommend letting AI access all that data. You can expose a small sample, to enjoy the benefits of visibility, while keeping the full details as your intellectual property.
Or you can adopt “a timed approach”. For example, if you have an exclusive story, you can restrict it for 3 days to get the traffic benefits, then open it up to AI crawlers so you enjoy the long-term brand visibility and authority benefits.
Managing AI is not “all or nothing”. For me it is about finding the right balance for brand relevance, not for traffic. But completely blocking AI crawling means losing visibility to a competitor.
You talk about crawling as a brand management tool, no longer just about indexing. But if a company’s reputation depends on the signals collected and reworked by AI systems, isn’t there a risk of handing your identity over to opaque and unpredictable systems? How can a brand build and defend its narrative in an ecosystem where the citation is worth more than the click?
AI systems look for consensus. Think of word of mouth. If you are FIAT, you can say you are fuel-efficient and affordable, and the broad landscape of media, reviews, forum threads, social posts, award coverage and certification bodies will largely agree. So LLMs will repeat that positioning.
If Ferrari tried to say the same, there would be no consensus in the corpus. And it is unlikely the LLM would “agree”.
To build a narrative, it is important to know your brand’s strengths and create authoritative layers of consensus. That is how you can consistently influence LLM outputs.
People keep talking about “quality content” as if it were the Holy Grail. You propose talking instead about “distribution-worthy content”, meaning texts that actually manage to reach and engage people. What changes, in practice, when you adopt this perspective? And why, in your view, is the obsession with EEAT leading many brands to lose visibility instead of gaining it?
An article can have “high EEAT”. It is well structured, written by experts, highlights lived experience with original images, sits on authoritative domains, and so on.
Yet it can still be unremarkable content. Something the target audience has already read many times. Something, frankly, boring. Distribution-worthy content will never be boring, because if it were, it simply would not get shared.
It is about recognizing that publishing the article is the starting point, not the end goal. You need to proactively plan which platforms you want to gain visibility on. If I want visibility in search, the way I craft the article is different from the one I would use if I wanted coverage on Google Discover.
Marketers often take a “they will come” attitude, because they have invested heavily in the content and believe quality is enough. It is not. For optimal visibility, content must drive engagement. To produce ROI, content must also increase brand relevance.
It is not about what you create. It is about who it reaches and whether it has an impact on the reader.
You describe the sitemap as a strategic tool to improve crawling and site visibility, not just a technical document to submit to Google. How can a “smart” sitemap really influence SEO and help brands understand where they are losing indexing or traffic?
When sitemaps are smartly split into sections, you can understand how Google evaluates the authority of a section. This is crucial, since search engines judge a domain based on everything that is indexed.
If you can easily identify that section A has 90% indexing while section B has only 15%, you know exactly where to focus your attention to improve the overall health of the site, and you can more easily measure the impact of SEO improvements.
Content that is not indexed effectively does not exist. What do you think is causing this growing “invisible index” of pages that Google and Bing do not register, and what are the most common mistakes that prevent content from truly entering the stage?
The vast majority of sites are bloated. They have thousands of low-value pages due to poor parameter handling, badly managed taxonomies, or the lack of page lifespan management.
Both Google and Bing are becoming more selective about what they allow into the index, since it is the corpus for AI-powered RAG surfaces. This means it has implications both for the quality of LLM output and for cost.
From a brand perspective, the existence of such pages slows down the crawling of important content, reduces domain authority and, for those that do make it into the index, delivers a horrible user experience that damages engagement.
Therefore, every page on the website should have a clear reason to exist. I always recommend using GSC with segmented sitemaps and regex on URL paths to analyze which sections are performing and which need improvement.
Finally, I would like to ask you something that really intrigued me… Marketers often look for inspiration in data or trends, but you suggest starting from complaints, that is, from the moments when the brand falls short of expectations. How can turning complaints into SEO actions truly strengthen a brand, instead of limiting them to simple “customer care” interventions?
Oh yes, complaints can be a gold mine! They offer useful insights to improve the message, fix weak points and act on the snippets that influence perception in search engines (and answer engines, editor’s note).
When the star rating of a product page drops below 4, CTR falls significantly. The same goes for generative models: if the information they absorb comes mostly from negative reviews, their answer will tend to advise against that brand. In this scenario, complaints are not just a customer service matter, but a direct indicator of how the brand is being represented.
I think we sometimes forget that marketing was built on the 4Ps, of which product was one. (The 4Ps model is a classic marketing framework describing the fundamental elements on which an effective strategy is built. The four components are: Product, Price, Place, Promotion, editor’s note).
Well, if the product is poor, no SEO will be able to make up for the negative perception in the long run.
Winning over AI agents: what is at stake for every brand
The interview with Jes Scholz highlighted an epochal shift: search is no longer a touchpoint between user and engine, but an ecosystem in which AI agents select, filter and decide on our behalf.
SEO remains central, but it is moving onto new ground, where code quality, brand strength, the consistency of external signals and the ability to be present before the search even begins are what count.
It is a change that forces brands to rethink their digital identity, because these intelligences evaluate what they find with no patience, no compromises, and without ever saying “oh well, come on, this will do…”.
No, AI systems do not settle for slow sites and confusing, unstructured pages!
From my point of view, the arrival of agents does not shrink the role of search professionals, it expands it.
It means learning to work for an audience that does not look at pages, but interprets them.
An audience that does not get emotional, but recognizes patterns.
An audience that, when it hits an obstacle, does not wait: it moves on.
Earning the trust of these systems requires technical rigor, genuinely distributed content and a brand capable of generating consensus in the information corpus AI draws from.
I thank Jes for her clarity, the depth of her insights and her willingness to engage in this conversation.
See you next week with a new episode of SEO Confidential!
#avantitutta
