+683% vs −67%: what separates those who grow from those who disappear in 2026 SEO

If you run a company and do not deal with digital marketing directly, start here.

Organic traffic, the people who find your company on Google without you paying to appear, is worth real, measurable money. The organic traffic of blissagency.it is currently worth €79.145 a month in equivalent Google Ads cost. This means that every month we do not spend that sum simply to exist online. We built it as an asset, rather than renting it.

Since the start of 2025, and more sharply after Google’s early-2026 update, the market has made its selection. Those who had built their online visibility the right way have kept growing. Those who had built it the wrong way, even after investing significant sums in previous years, have lost ground quickly and in ways that are hard to reverse. It is not a question of budget spent: it is a question of the architecture of the choices made.

There are three questions worth putting to your marketing lead or to the agency managing your digital presence. First: how many organic visits does the site generate each month, and what has the trend been over the last twelve months? Second: does the site appear in Google’s AI panels when someone searches for the services you offer? Third: is the content architecture designed to capture clients who are evaluating a purchase, or only to answer generic questions? If the answers to these three questions are not immediate and backed by data, the problem is structural, not tactical.

The rest of this article is written for those who manage SEO day to day. If you want to share it with your team or your communications partner, that is exactly what it was built for.

In the first article in this series we described how blissagency.it went from 3.442 to 26.974 monthly organic visits in twelve months. If you have not read it, start there. This is the sequel, and to understand it you need the context.

00 · BEFORE WE BEGIN // A QUESTION THE INDUSTRY IS NOT ASKING ITSELF

The SEO market is not in crisis. It is selecting.

blissagency.it domain trend, 24 months: from almost zero to 26.974 monthly sessions. The curve on the right is the one the market did not expect.

Charles Darwin never said that the strongest survive. Popular culture says it, having oversimplified one of the most precise scientific observations ever made. What Darwin actually described is subtler, and more useful: those who survive are the ones best suited to the conditions that exist now. Not those best suited to yesterday’s. Not those who have accumulated the most resources over time. Those who adapt to the current context, and do so quickly enough not to become extinct in the middle of the change.

Nassim Taleb formalised a complementary principle with the Lindy Effect: what has stood the test of time is statistically more likely to keep standing. A book that has lasted two hundred years is more likely to last another two hundred than a book published last week. But Taleb adds a caveat that is often forgotten: the Lindy Effect applies to ideas and systems that are not fragile to changes in context. A company that has prospered for thirty years in a stable market is not automatically antifragile; it has simply survived a context that never truly changed the rules of the game.

The SEO market has changed the rules of the game. Not gradually. Not with an update that penalised some and rewarded others, as has happened cyclically since Google has existed. It has changed the underlying model, the way information is extracted, synthesised and returned to users. And this kind of change does not forgive those too rigidly structured to adapt.

Sam Altman has, on several public occasions, distinguished between two categories of tech companies: those that build their products on the AI capabilities available today, and those that design on the assumption that AI will be significantly more powerful in the short term. The second category, Altman argues, has a structural advantage that grows over time, because every improvement in the models amplifies a foundation already built to accommodate it, rather than requiring a complete re-architecture.

The data that follow are not projections. They are measurements. And what they measure is a selection under way.

The question is not whether the SEO market is changing. It is whether the structure you have built is designed to survive the change, or for the market that existed when you built it.

01 · THE MARKET // THE NUMBERS NOBODY WANTS TO READ

Since early 2025, sixteen months ago now, organic traffic across the Italian web has been falling. Systematically. And it has not recovered.

There are ways of describing what has happened to organic traffic in Italy since the start of 2025, and what is still happening as we write, in May 2026, that sound reassuring. You can call it a “transition period”. You can talk about an “algorithm update” and imply that things will soon settle down. You can point to seasonality. You can say that “the long tail is holding up” and that “anyone with quality content has nothing to fear”.

We prefer data.

The data show that from late 2024 onwards a systematic contraction has hit, and sixteen months on is still hitting, sites built on the dominant model of the past ten years: large volumes of content, broad keyword coverage, aggressive link building, classic on-page optimisation. That model worked superbly for a decade. It stopped working the moment Google stopped being primarily an index of links and became primarily a direct-answer system.

It is not a content quality problem. It is not a backlink problem. It is an architecture problem, and architectures are not changed in a quarter.

The problem is not that Google has penalised anyone. It is that Google no longer needs to send traffic to those who merely act as intermediaries between the question and the answer.

02 · THE NATIONAL PRESS // THE PARADOX THAT SAYS IT ALL

Google cites them more and more. And sends them fewer and fewer visitors. It is not a paradox; it is the new model.

ilmessaggero.it: organic traffic −508.207 sessions, AI Overview keywords +32.410. Google cites it more and visits it less. 

lastampa.it: the same dynamic. Traffic −157.583, AI Overview keywords +1.424. Editorial authority no longer protects traffic.

If you want to understand what has happened to organic traffic over the past year and a half, the best place to look is not marketing agencies. It is national news outlets. Because news outlets have everything that should, in theory, protect you from algorithmic entropy: very high domain authority, backlinks from hundreds of thousands of sites, structured editorial teams, decades of digital history, unquestionable brand recognition.

And yet SEOZoom data from the last eighteen months tells a precise and unequivocal story.

Source: SEOZoom Website Analysis · May 2026

Let us pause for a moment on the Il Messaggero figure. Organic traffic fell by more than half a million monthly sessions. Over the same period, keywords cited in AI Overview grew by 32,410. Google cites Il Messaggero as a source in over 32,000 more queries than a year ago. And sends fewer visitors to the site.

La Stampa: +1,424 keywords in AI Overview. Organic traffic: −157,583. Wikipedia: 346,916 keywords in Google’s AI panels, one of the widest footprints in the entire Italian publishing landscape. Organic traffic: down 44 million monthly sessions.

This is not a figure that can be explained away by “low-quality content”. Wikipedia does not have an editorial quality problem. Il Messaggero and La Stampa are publications with professional newsrooms, established editorial processes and recognisable bylines. The problem is not what they publish. It is the model through which Google serves up what they publish.

Google has stopped acting as an intermediary. For informational queries, which accounted for the largest share of organic traffic on any editorial site, Google answers directly, extracting information from the domains it considers primary sources and returning it within the interface without requiring a click. Newspapers are used as invisible infrastructure. They are cited. They are not visited.

This is the mechanism that explains everything else.

Being cited by Google’s AI Overview means you have built real semantic authority. It also means that this authority, unless it is paired with transactional and commercial search intent, no longer generates traffic. Google uses you without sending visitors back.

03 · AGENCIES // THOSE WHO FAIL TO ANTICIPATE THE MARKET PAY THE HIGHEST PRICE

The content→volume→traffic model has stopped working. Those who failed to adapt in time are now counting the damage.

One of the most structured marketing agencies in Italy: a peak of ~70,000 sessions in 2024, 23,490 today. −67% in eighteen months.

If the mechanism is the one described above, with Google answering informational queries directly and replacing the traffic to source sites, then the most exposed category is not news outlets. It is sites built mainly on informational content optimised for medium-to-high-volume keywords. Sites that have spent years producing articles, guides, comparisons and structured answers designed to be found in informational SERPs.

Among Italian marketing agencies, this model was dominant. It made sense: publishing informative content on marketing, branding and digital advertising brought qualified traffic, demonstrated expertise and generated natural backlinks. It was the right strategy for the 2018-2023 market.

The 2025-2026 data show what happens when the market changes faster than the architecture you have built.

Source: SEOZoom Website Analysis + Competitor Analysis · May 2026. Peaks estimated from the 24-month SEOZoom chart curves.

The pattern is clear. The most severe contractions hit the sites most fully built on the content-volume model: thousands of ranking keywords, years of editorial output, and traffic that seemed defensible precisely because it was spread across such a broad base. The broad base turned out to be a weakness, not a protection. More surface, more exposure to changes in the underlying model.

This is not about pointing fingers. The choices made in those years were rational, consistent with the evidence available, and probably delivered excellent results up to 2024. The problem is not having got it wrong. The problem is failing to adapt quickly enough, because a structure built on thousands of pages of informational content cannot be re-architected in a quarter.

The SEOZoom competitive map for blissagency.it in May 2026: Bliss in fourth place overall for traffic, ahead of every competitor that ranked above it a year ago.

This is the difference between fragility and antifragility. Fragile systems hold up to a threshold, then break. Antifragile systems grow stronger under pressure. The question is: which of the two have you built?

The losers were not those who did things worse. They were those who had built on a model the market stopped rewarding. It is an important distinction, because the problem is not operational, it is strategic. And strategic problems require structural solutions, not tactical adjustments.

04 · THE CAUSE // WHY THIS IS NOT AN ALGORITHM UPDATE

Algorithm updates have always had a pendulum effect. Not this one. And the early 2026 update settled the debate.

“Earthquake in Google’s SERPs”: the Italian SEO community in March 2026. Not a feeling, a shared measurement.

Link to the full article: https://www.roberto-serra.com/news/volatilita-serp-google-aggiornamento-11-3-2026/

Anyone who has worked in SEO for more than five years has learnt to live with Google’s algorithm updates. The cycle has always been recognisable: Google updates, one category of sites loses traffic, another gains it, the market realigns within a few months, the new rules of the game are studied, the strategy is adapted. It happened with Penguin, with Panda, with the Core Updates, with the Helpful Content Update. Each time someone lost, someone else gained, and total web traffic remained broadly stable.

The contraction that began in late 2024 was definitively confirmed between January and April 2026. In those four months Google released a broad, deep update, documented and discussed by SEO specialists across every digital channel, which reshuffled the SERPs sharply and lastingly. This was not yet another fine-tuning correction. It was a structural redesign of ranking priorities, with visible, measurable effects on sites depending on whether or not they had anticipated the direction. That update produced winners and losers: the traffic curves we show in this article are direct evidence of it.

What began to happen at the start of 2025, and which by May 2026 had shown no sign of reversing but had instead consolidated, is unlike anything that came before. Not in intensity, but in nature.

Algorithm updates used to redistribute traffic. AI Overview reduces it. It does not redistribute it across different sites; it keeps it within the Google ecosystem. When a query is answered directly in the AI panel, the click does not go to the site Google judged best: it goes nowhere. Google has answered directly. The user got what they were looking for without leaving Google.com.

This mechanism was not present in previous cycles. Google’s updates always assumed that the definitive answer lay on an external site; the goal was to find the best site and send the user there. AI Overview changes this premise: for a growing share of queries, Google considers that it can answer directly, without intermediation.

The system will not realign the way it always has. Because this is not about changing the criteria for deciding who deserves traffic; it is about changing how much traffic is passed to external sites at all. And that figure will not return to pre-2025 levels. It will move in the opposite direction as Google extends AI Overview coverage to ever broader categories of query.

The most attentive research agencies estimate that informational queries with high AI Overview coverage have already cut organic CTR by more than 30% for positions 1-5 in the affected SERPs. It is not an estimate; it can be measured in the Google Search Console of any site that has tracked them over time.

Do not wait for the market to realign as it always has. This time it is not realigning, it is being redesigned. And the new design rewards those who have built on transactional and commercial search intent, not on informational volume.

05 · THE NEW ECOSYSTEM // THREE ENVIRONMENTS, THREE LOGICS

Google, ChatGPT, Perplexity. They are not variants of the same channel. They are ecosystems with distinct logics.

One of the most common misreadings at the moment is treating Google’s AI Overview, ChatGPT, Perplexity and the other AI engines as variants of the same phenomenon. They are not. They have different citation logics, different source-selection mechanisms, and different implications for anyone who wants to be present in the answers they generate.

Google AI Overview operates within traditional SERPs. It selects sources mainly on the basis of classic SEO signals (domain authority, semantic relevance, content structure, markup) but filters them through a citability assessment layer: ranking is not enough; content must be structured so that it can be extracted and synthesised. Sites with correct JSON-LD markup, an explicit question-and-answer structure, FAQ schema and consistent hierarchical headings are favoured as sources. The rest are ignored, even when they hold first organic position.

ChatGPT and the new generation of language models work on a different logic. They do not index the web in real time as Google does. They build their knowledge during training, updated periodically, and add real-time search for queries that need recent data. Their citation mechanism is driven by semantic reputation: how often a brand, a domain or an author is mentioned positively and authoritatively in the body of text the model has ingested. Appearing in Google’s AI Overviews helps build this reputation. Having content cited by other authoritative sites helps even more.

Perplexity works in a hybrid way: it indexes the web in real time and produces synthesised answers with explicit citations. It is the channel where well-structured, recent content delivers the most direct and measurable results in terms of citation.

The practical consequence is that optimising for a single ecosystem is not enough. A site can rank extremely well in traditional Google SERPs and be completely absent from AI answers. Or the reverse: it can be cited frequently by ChatGPT and Perplexity while having modest organic traffic in classic SERPs. Positioning across the three environments requires partly distinct strategies, but with a common foundation: genuine semantic authority, content structured to be extracted, and declarative markup that tells machines what you are, what you do and who you are.

Traffic from AI engines is not yet the main channel for most sites. But it grows every month, with every new user who replaces a Google search with a question to ChatGPT. Those who start building authority in this ecosystem today will, in twelve months’ time, have a structural advantage that will be almost impossible to close.

05b · INFORMATIONAL SERPS // GOOGLE IS ACTIVELY LOOKING FOR A SUBSTITUTE

For informational queries, Google is not just adding AI Overviews. It is shifting the first page towards YouTube videos.

Incognito search for “come fare branding” (how to do branding), May 2026: the top three positions are YouTube videos. Text articles start further down.

There is a signal that those who monitor SERPs systematically have been observing with increasing frequency since early 2026: for a significant share of informational queries, “how X works”, “what Y means”, “how to do Z”, the top organic positions are no longer held by blog articles or editorial pages. They are held by YouTube videos, often accompanied by video carousels covering half of the visible page above the fold.

This is not a coincidence. It is a choice in SERP architecture. For informational queries where the AI Overview already provides a concise answer, Google has a clear incentive not to send traffic to competing text-based sites: it prefers to answer directly with AI, and for the visual and in-depth component it prefers YouTube, which it owns. Traffic stays within the Google ecosystem. The click, when there is one, goes to a platform Google controls directly.

The operational implications are direct. A site that produces only text content optimised for informational queries finds itself competing on two fronts at once: the AI Overview that answers without sending clicks, and the YouTube videos that occupy the traditional organic positions. For those without an integrated video content strategy, this is territory that is closing structurally, regardless of the quality of the articles produced.

The answer is not to stop producing text content; it is to stop producing it for purely informational queries that carry no transactional or commercial intent. The text that still holds value is the kind that answers the questions a prospective client asks while weighing a purchase decision, not while looking up the definition of a term.

Google is actively replacing informational traffic to third-party sites with AI answers and YouTube videos. This is not an emerging trend; it is a declared policy, expressed through its SERP layout choices. Anyone who has built their traffic on purely informational queries with no commercial intent is building on a foundation that Google is progressively removing.

+683% in twelve months. In the same period in which the market was falling.

blissagency.it in May 2026: 26,974 organic sessions, traffic value €79,145/month, 336 keywords in AI Overviews, 414 referring domains.

In the period in which sito1.com lost 67% of its organic traffic and sito2.com 63%, blissagency.it grew by 683%. Not in absolute terms, but relative to itself. From 3,442 monthly organic sessions to 26,974. With Zoom Authority rising from 32 to 48. With 414 referring domains. With 336 keywords appearing in Google’s AI Overviews.

Source: SEOZoom Website Analysis, blissagency.it · May 2026

The year-on-year figure calls for careful reading. A year ago blissagency.it generated 3,442 organic visits a month. The SEOZoom annual projection shows that over the previous three months traffic grew from 17,261 to 26,974, almost 10,000 additional sessions in a single quarter. The curve is not linear: it is exponential. Exactly as we described in the first article of this series when we discussed the compounding effect of SEO.

In the organic competitor table extracted from our SEOZoom dashboard, blissagency.it ranks fourth overall for traffic in the reference group, with 26,974 sessions, ahead of NetStrategy (23,490), SayAgency (16,160), Dexanet (11,620) and SecretKey (5,196). Twelve months ago, every one of these sites ranked above us.

This is not luck. It is not seasonality. It is not an update that happened to reward us while penalising others. It is the result of architectural choices made ahead of the shift in the market, and of a system that keeps growing stronger while others erode.

Bliss did not benefit from its competitors’ decline. It built something that works in the new market, while those who had built for the old market were losing ground. These are two parallel stories, not one the consequence of the other.

07 · THE CHOICES // WHAT WE DID DIFFERENTLY

Not more content. Better content, within a structure designed for the market that exists now.

The wrong response to the contraction of informational organic traffic is to produce more content. It is the instinctive response, and it is wrong for the same reason that adding more fuel to a poorly running engine does not make it go faster. The problem is not the quantity of input; it is the architecture of the system.

What we did on blissagency.it was not to publish more. It was to build differently. With a few precise choices that can be described, measured and replicated.

1 · Architecture by intent, not by volume

The central strategic decision was to assign each level of the site a precise, non-overlapping search intent. Homepage: navigational and branded intent. Service pages: commercial and transactional intent, from people looking for an agency and considering getting in touch. Blog: informational intent on topics where Bliss’s authority is relevant and defensible over the medium term.

This means that every keyword with commercial value lives on a single page of the site, designed specifically for that intent. Not on four pages cannibalising one another. Not spread across informational articles that dilute the signal. Concentrated, declared, defensible.

Keyword cannibalisation is one of the most underestimated SEO problems because it produces no visible errors. The site works, the pages get indexed, the content looks fine. But faced with four pages answering the same query with overlapping intents, Google does not know which to reward, and often rewards none. The cannibalisation audit completed on blissagency.it identified and resolved overlaps across all the main keywords. The result was visible in the following weeks, without publishing a single new piece of content.

2 · JSON-LD and semantic markup: Google does not optimise pages, it recognises entities

Google Rich Results Test on blissagency.it: 19 valid items detected. Articles, Breadcrumbs, FAQ, Local Businesses, Organisation, every entity declared and verified.

There is a paradigm shift in how Google evaluates and indexes sites that most people have not yet taken on board. For years SEO was conceived as the optimisation of individual pages: each URL had its own keywords, its own title tag, its own meta description. One page, one objective, one ranking. This model still holds for the basic technical layer, but it is no longer enough to describe how Google’s AI systems build their understanding of a domain.

Google no longer reads a site as a collection of pages. It reads it as a graph of entities, organisations, people, places, products, concepts, with declared relationships between them. A communications agency is not “a page that talks about marketing services”: it is an Organization entity with a verifiable headquarters, identifiable founders, declared services, documented clients and an editorial history attributable to real authors. This is the difference between a site Google understands and a site Google attempts to interpret, with all the uncertainty that entails.

AI systems, both Google’s AI Overview and language models such as ChatGPT and Perplexity, select the sources they cite based on the clarity and consistency of this semantic graph. A domain that explicitly declares its entities and the relationships between them, through correct and complete JSON-LD markup with an @graph structure, gives these systems a readable and verifiable map. A domain that does not is interpreted heuristically, with all the margins of error and omissions that follow.

On blissagency.it we have implemented a JSON-LD @graph structure across the entire site: Organization, WebSite, WebPage, BreadcrumbList, Article, FAQPage, Service, with every entity declared with explicit relationships between them, not as isolated blocks. The result is not only technical: it is semantic. Google knows precisely who Bliss is, what it does, where it is based, who works there, which clients it has served, what content it has produced and who wrote it. The 336 keywords appearing in Google’s AI Overviews are no accident. They are the direct result of a domain that speaks the language of machines with a structured vocabulary, not of pages optimised for individual keywords.

3 · E-E-A-T as a verifiable system, not a statement, including author diversity

Andrea Maso’s author page on blissagency.it: a verifiable biography, declared expertise, more than 190 published articles. Not a name, but an entity that AI systems can recognise.

Experience, Expertise, Authoritativeness, Trustworthiness. Google made its editorial quality criteria explicit with this acronym, and ever since every SEO agency has talked about E-E-A-T as if it were a checklist of things to put on the page. It is not a checklist. It is a system of verifiable signals that Google assesses in the context of everything it can know about a site, an author, an organisation.

An often overlooked element is the variety of authors within a site. A blog always signed by the same author, however excellent, generates a one-dimensional authority signal. A site where several professionals with distinct expertise contribute under their own name, each with a verifiable profile and a declared area of expertise, builds an authorship graph that Google reads as an indicator of a real organisation and distributed expertise. It is not a tactic; it is the digital reflection of how a serious professional organisation works.

On the blissagency.it blog, content is signed by distinct authors, Andrea Maso, Francesco Acri, Luca Maletta, Gianluca Sabbini, Mario Antonini, each with a verifiable profile page, documented experience and a public biography. They are not fictitious authors created for SEO purposes: they are the people who actually work on client projects. This consistency between who signs the content and who does the work is one of the strongest signals of genuine E-E-A-T, because Google can verify it independently through LinkedIn, external publications and editorial mentions.

On Trustworthiness: complete, verifiable contact details, a public VAT number, real physical addresses at both offices in Rome and Milan, declared ISO certifications. Not as a formality, but to build trust signals that automated systems can verify independently.

4 · Continuous technical audit as a practice, not a one-off intervention

A WordPress site built with Elementor accumulates technical SEO errors over time, silently, without anyone noticing until they show up in the traffic. Multiple H1s generated by Elementor widgets that the theme does not recognise as headings but that crawlers read as such. Missing meta descriptions on automatically created URLs. Duplicate title tags on CMS-generated pages. Podcast CPTs not excluded from the sitemap, diluting the crawl budget. Orphan pages with no internal links, receiving no domain authority.

In this working session we produced four technical remediation files: 30 URLs with missing meta descriptions, 3 pages with duplicate title tags, 8 orphan pages with an action plan, and 19 pages with multiple H1s, each diagnosed with its root cause and a specific fix. These are not small numbers for a site of moderate size. They are errors that build up over time and that every crawler reads as negative signals about the site’s technical consistency.

A technical audit is not something you do once a year. It is an ongoing practice, exactly like maintaining a production plant. A site that is not technically audited every quarter accumulates entropy. A site that is audited systematically stays within the optimal range of readability for crawlers.

None of these four choices is new. They have been established practice in SEO for years. The difference is not knowing what needs to be done; it is having the discipline to do it systematically, methodically, without stopping when results are not yet visible. The compounding effect works only if the system runs continuously.

08 · DARWIN AND LINDY, THE FRAME THAT EXPLAINS EVERYTHING ELSE

It is not the strongest who survive. It is those structured to adapt to the conditions that exist now.

Let us return to where we began. Darwin, Lindy, and the question this article has been trying to answer from the start.

The national newspapers losing traffic have not become less authoritative. Wikipedia has not made its content worse. Il Messaggero and La Stampa have professional newsrooms, decades of history and backlinks from half the world. The problem is not quality. The problem is that their structure, an editorial model based on publishing volume, monetisation tied to advertising traffic, a newsroom organised to maximise output, is not designed to adapt quickly to a change in how traffic is distributed. They have too many stakeholders, too many entrenched processes, too much structure to turn in time.

This is precisely the cost of inertia that Taleb describes when he talks about fragile systems. Fragile systems do not break because they are weak; they break because they lack the flexibility to absorb shocks. A robust system withstands shocks. An antifragile system grows stronger under pressure. A fragile system collapses.

The Lindy Effect, applied to SEO, does not say that whoever has the longest digital history will survive. It says that those who have built on principles that do not depend on short-term market conditions, real semantic authority, declarative structure readable by machines, search intent defensible over the long term, have firmer foundations than those who have built on tactics optimised for the rules of the moment. Tactics change every time Google updates its algorithm. Principles endure.

Darwin completes the picture. Selection is not cruel by choice; it is neutral by nature. The SEO market is not punishing anyone. It is simply ceasing to reward what no longer fits the current context. Those who adapt survive and thrive. Those who fail to adapt, through inertia, through structure, through an inability to recognise change in time, are replaced by those who have adapted.

Right now, in the Italian SEO market for communications agencies, the selection is already under way. The data in this article document it. Who is gaining traffic and who is losing it can be measured precisely enough to draw operational conclusions.

The question that remains, and one that every CMO, every marketing director, every business owner who depends on organic traffic should be asking, is not whether the change is happening. It is whether the structure they have built is designed to adapt, or for the market that existed when they built it.

Darwinian selection does not wait for you to be ready. When the data show that the context has changed, the useful response is not to wait for further confirmation; it is to adapt the structure before the cost of inertia becomes unsustainable.

09 · THE CONCLUSION, ONE THING TO TAKE AWAY

Organic traffic will not come back on its own. Either you build it for the market that exists, or you lose it to the one that no longer does.

This article is long. It is long because the phenomenon it describes is complex, and simplifying it would make it useless. But if we had to take just one thing away from this analysis, it would be this.

The business model of informational organic traffic, lots of content, broad keyword coverage, optimisation for volume, became structurally fragile the moment Google began answering informational queries directly. This is not reversible. Do not wait for Google to backtrack on AI Overviews. Do not wait for the next update to “rebalance” the SERPs. The change is structural, not cyclical.

What survives, and what the blissagency.it data confirms with 26,974 monthly sessions in a declining market, is a different approach. Content that answers search intent that is defensible over the long term. Declarative architecture that speaks the language of machines. Semantic authority built on verifiable sources. Technical audits as systematic practice. And the discipline to sustain these things over time, even when the results are not yet visible.

If your site’s structure was designed for the 2023-2024 SEO market, the time to rethink it is now, not after the data becomes impossible to ignore.

Frequently Asked Questions

What is Google’s AI Overview and why does it affect organic traffic?

AI Overview is the direct-answer panel that Google displays at the top of the SERP for a growing share of queries, mainly informational ones. The mechanism extracts information from selected sources, synthesises it and returns it to users without them having to click through to any external site. The result is a drop in organic CTR for traditional positions 1-10, particularly severe for high-volume informational queries.

Is the decline in organic traffic the same for every site?

No. Sites built mainly on informational content optimised for medium-to-high-volume keywords are the most exposed. Sites with a strong commercial and transactional intent component, a semantic architecture structured to be cited by AI systems, and verifiable editorial authority show very different trends, in some cases significant growth even in the current period.

What can be done to appear in Google’s AI Overviews?

The most significant factors, based on an analysis of the sites that appear most frequently in AI Overviews, are: complete and correct JSON-LD markup, a content structure with explicit questions and structured answers, an implemented FAQPage schema, domain authority on clearly defined and declared topics, and primary sources cited within the content. There is no guarantee of being cited, but the absence of these elements is often correlated with absence from AI Overviews.

Is traffic from ChatGPT and Perplexity already measurable?

Yes, through platforms such as SEOZoom AI Analytics. Traffic from AI engines still accounts for a minority share of total traffic for most sites, but it grows every month. blissagency.it records sessions from ChatGPT, Perplexity, Claude and Gemini, a category of traffic that did not exist eighteen months ago. The right strategy is to build authority in this ecosystem now, not once the traffic has already become significant for competitors.

How long does a strategic SEO repositioning take?

It depends on how deep the change needs to go. Technical work, such as audits, fixing on-page errors and markup, produces results within weeks. Re-architecting content and the structure of topic clusters takes three to six months to complete, and as long again to be assessed by crawlers. A full repositioning, from diagnosis to consolidated results, rarely takes less than twelve months. The reason the right time to start is now, and not in six months, is that every month of delay is a month in which the gap with those who adapted earlier keeps widening.