Written by Juan Antonio Castillejos
Index
On July 25, our head of SEO María Navarro gave a talk with the same title as this article at SEOplus, an SEO conference organized by Luis M. Villanueva and Álvaro Sáez in Alicante. Today we’re sharing a recap of her presentation in case you couldn’t attend.
The fight for visibility has been, ever since SEO has been SEO, the cornerstone of any online project or business that wants a real shot against its competition.
As Google’s algorithm—the world’s leading search engine—has evolved, that visibility isn’t limited to text content anymore; it also includes videos, podcasts, and social posts, depending on search intent.
On the other hand, being considered by Google as qualified, recognized experts in our category or specialty (the oft-cited E-E-A-T) is now essential for a company, project, or professional to even have a chance of being discoverable in the ocean of results.
If we add to this scenario the arrival of AI platforms as new agents producing answers to certain user queries, the fight to be recognizable and visible promises to be even more complex. Sharing information no longer guarantees traffic.
Reality check: the data
The talk kicked off with a look at the traffic figures that, in mid-2025, the various platforms through which a user can conduct a search are handling.
The takeaway is that Google is still the user’s preferred way to search, with nearly 140 billion monthly visits, according to a study by Exploding Topics.
The second most visited platform would be YouTube (owned by Google), with 78.6 billion monthly visits. Finally, ChatGPT currently receives around 5.2 billion.
Google would be receiving roughly 27 times the monthly visits that ChatGPT gets, so the gap is still huge.
Even so, as we saw earlier, these searches no longer guarantee traffic from Google to the websites that create the content.
The rollout of AI Overviews in Spain on March 30, 2025, brought a paradigm shift in the direct relationship between impressions and clicks—in other words, between being seen and getting traffic.
Now, your content may be used to answer the user but not attract traffic directly to your website.
That’s the new reality of SEO.
Your content is visible, but not as visitable
AI Overviews—the AI-generated results in Google Search—are shown to users in 42.5% of searches, which translates into a general decline in click-through rate and a direct impact on traffic, more or less significant depending on the vertical.
Within those figures, 96% of the times AIO appears are to answer queries with informational intent.
In other words, generative AI answers appear almost universally when the user query is informational (“what is…,” “how to…”). However, they show up less often for commercial searches (“alternatives to iPhone 16) and practically never for transactional searches (“buy iPhone 16”). In the latter case, Google prioritizes product listings, ads, or e-commerce results.
Sharing information no longer guarantees traffic: is it a good idea to stop investing in content?
Informational content drives less traffic than before. That’s the reality for all portals and websites. Generative AI answers are satisfying a large share of users’ search intent.
Before deciding where to allocate resources, we need to analyze the following:
What common characteristic do searches have when they can be fully satisfied by automated answers?
In the talk, we distinguished between searches that can be satisfied with direct answers and searches for which, by their nature, the user needs to dig deeper.
Imagine we manage the content for a farming blog. Our post “When to Plant Tomatoes – Step-by-Step Guide” will likely see its traffic decrease for the query “best time to plant tomatoes” because AI Overviews provides a direct answer. In this case, AIO gives the key time window. The average user doesn’t need more technical detail.
Therefore, the searches most affected by generative AI answers will be those that can be addressed with a simple, direct response.
But what happens when the user is looking for information on important topics that can have a bigger impact on their day-to-day life?
In those cases, the user will very likely need to dive deeper in their process of seeking information, data, and opinions. Not only will the first answer from generative AI be insufficient; they’ll also need to consult reliable and expert sources to fully satisfy their information need.
As María Navarro put it, AI answers can be useful, but when it comes to important decisions the user needs more: to explore, compare, go deeper, and validate.
Here are two examples from the talk:
- “Best cream for sensitive skin with rosacea in summer”: the user will dig deeper because they need specific brands, real opinions, prices, and availability.
- “How to reduce personal income tax if you’re self-employed”: the user is looking for examples, exact thresholds, particular cases, and confirmation from official sources.
If users will still need expert advice for almost any topic or specialty that can’t be solved with a direct answer, do we want to work to be part of that answer even if, overall, we get less informational traffic?
The right answer for a project seeking visibility should be yes.
The winning strategy would be to identify where the user’s “deeper” doubts lie—the ones AI can’t satisfy—and address them with our resources. This content should be characterized by its depth, authority, trustworthiness, human value, and the use of first-party data.
How to create relevant content AI can’t answer
María Navarro’s talk continued with a review of the key points for a user-relevant content strategy that generative AI cannot address because it requires human experience and value:
Classify informational answers as quick or complex
Analyzing user search intent is no longer enough; we need to go a step further: identify whether the query is solved with a “quick” or a “complex” answer—that is, whether it requires a higher degree of investigation on the user’s part.
Take the term “cortisol” as an example.
A query related to its definition (“what it is”) would be a quick answer, handled by AI Overviews.
But beneath that definition-level query, others emerge that require greater depth, knowledge, and expertise. For example: “how to lower high cortisol” or “how high cortisol affects my health.”
Create unique, in-depth content with real differentiating value
When a query calls for nuance, real-world examples, testimonials, advice based on the source’s experience, comprehensive guides, comparisons, or data… the real opportunity to stand out is to produce that unique, differentiated content that positions the project as a reference on the topic. We probably won’t achieve it with a single piece, but by accumulating distinctive, high-quality content across different areas in topic clusters that signal to Google our areas of expertise.
What can generative AI not offer that a writer or subject-matter expert can?
- Human perspective and first-hand experience.
- Real narrative, with up-to-date context.
- Practical guides based on concrete cases.
- Useful, up-to-date data versus generic data.
- Honest comparisons with expert opinions.
- Testimonials and opinions from real users that create connection.
E-E-A-T: an old friend, more relevant than ever
When creating content, we must demonstrate experience, expertise, authority, and trust.
To do so, we need to assign expert authorship in the field, with a demonstrable, verifiable track record, and use precise language in the content itself.
Accompany the user through the full journey
While many user searches will start by discovering the definition of a concept, it’s very likely that the user’s journey will continue with successive queries that approach that concept from different angles and perspectives.
Therefore, we have to develop content that digs deeper into the idea to accompany them throughout their research.
When the user wants to know what cortisol is, we should have a resource that addresses this query. And even if generative AI provides a direct answer, if we’ve built strong content and E-E-A-T signals, our content is likely to be used to form the automated answer and thus be cited as a reference source.
Sooner or later, that query will be followed by others we can target with content that requires significant amounts of experience, expertise, and authority, such as:
- What are the symptoms of high cortisol?
- Causes and treatments for high cortisol
- What’s the relationship between cortisol and anxiety?
- Which natural supplements help reduce cortisol?
- Does ashwagandha block cortisol?
Beyond creating one-off articles, your strategy should aim to develop strong topical pillars, build satellite content around them that addresses all of the user’s real questions, and interlink them.
Aim to be cited or linked as a relevant content source
Generative AI and rich panels don’t produce answers out of thin air; they draw from sources Google considers trustworthy, specialized, and relevant.
Therefore, a website’s goal should be to become a quality source for Google, so that other sites want to link to its content.
Monitor and adapt your results
The SERP has evolved and will keep evolving. We need to constantly check the performance of our assets and any shifts in the result types Google serves for the main queries that make up the project or business’s ecosystem.
Based on this analysis, we must be able to adapt our content to the form and format Google requires at any given moment. To adapt is to endure.
In short, SEO in 2025 is no longer just about creating information and waiting for traffic, but about understanding which questions AI can answer and which still require human experience, authority, and voice. The key is to design strategies that deliver depth, context, and credibility, building solid topical coverage and accompanying the user at every stage of their search.
Only then will projects maintain visibility, be recognized as reliable sources, and ultimately continue to play a meaningful role in an environment dominated by algorithms, generative AIs, and automated answers.
