SEO Is Not Dead — It’s Being Reweighted. Here’s What the Research Actually Shows

Introduction: Everyone Has an Opinion. Now There Is Research. SEO is Not Dead!

If you have spent any time in digital marketing in the past two years, you have heard the claim. Sometimes it is dramatic: SEO is dead, AI has killed it. Sometimes it is more nuanced but ultimately reaches the same conclusion: backlinks no longer matter, Google rankings are irrelevant, the only thing that counts now is getting cited by ChatGPT.

On the other side, there are the voices insisting that nothing fundamental has changed. Keep building links, keep publishing content, keep optimising for Google — the AI search thing is noise and will settle down.

Both positions are wrong — and the question of whether SEO is dead now has a research-grounded answer. In 2026, we finally have a study rigorous enough to say that with confidence.

In March 2026, independent researcher Dmitry Kargaev published a cross-paradigm comparative study titled “The SEO-to-GEO Gap: Quantifying Ranking Factor Divergence Between Traditional and Generative Search” on SSRN. It is the first study to directly compare SEO and GEO ranking factors using a normalised evidence framework — a Divergence Index that translates incompatible metrics from different studies into a comparable scale. The paper draws on a carefully triaged evidence base that includes Aggarwal et al.’s landmark GEO benchmark from KDD ’24, Ahrefs’ large-scale AI brand visibility study across 75,000 brands, Backlinko’s analysis of 11.8 million Google search results, and Semrush’s 2024 ranking factors study across 16,000+ keywords.

The conclusion is neither the dramatic death narrative nor the complacent continuity narrative. It is more precise — and more actionable — than either.

The answer to “is SEO dead?” is: no. It is being reweighted.

This post unpacks exactly what that means: which signals persist, which signals are newly important, which popular claims the evidence does not support, and what the practical implications are for businesses trying to be visible in a search landscape that now includes both Google rankings and AI-generated answers.

Why the “Is SEO Dead?” Question Keeps Returning

The question “is SEO dead?” has teeth because the surface evidence is genuinely alarming if you look at it quickly and stop there.

Google AI Overviews — the AI-generated summaries that appear above all organic results — now appear across a meaningful and growing share of queries. BrightEdge research from 2025 documented that Google search usage actually increased by 49% following the rollout of AI Overviews, but simultaneously reported that click-through behaviour for certain query classes has weakened as answer surfaces become more self-contained (BrightEdge, 2025a; BrightEdge, 2025b). Users are getting answers from the AI summary without clicking through to the ranked pages beneath it. For publishers and businesses that built their traffic models on organic click volumes, this is a real and material change.

Alongside the AI Overviews story, there is the generative search story. ChatGPT surpassed 800 million weekly active users in 2026. Perplexity has become a default research tool for professionals. Microsoft Copilot is embedded in enterprise productivity software used by corporate decision-makers daily. These platforms do not present ranked link lists. They present synthesised answers, and the sources they cite are not necessarily the sources that rank on page one of Google. Research by Kargaev (2026) notes that 72% of URLs cited in AI-generated responses do not rank in Google’s top 100 — a striking illustration of the divergence between traditional and generative search visibility.

So the question is legitimate. Something has changed. The issue is that the popular responses to what has changed tend to dramatically overstate the break while underspecifying what actually replaces what.

Kargaev’s paper identifies the core methodological problem precisely: SEO and GEO studies do not measure identical constructs. Traditional SEO research reports correlations with rank position. GEO research reports visibility gains from interventions, mention frequency across AI systems, or overlap between AI citations and already-ranked URLs. Without a framework that explicitly separates these constructs and normalises across them, comparing SEO and GEO becomes a comparison of unlike things — and the conclusions that follow are unreliable. Most of the “SEO is dead” narrative is built on exactly this kind of conflated comparison.

What the Research Actually Measured

Before drawing on the paper’s conclusions, it is worth understanding what it actually did — because the methodology is what gives the findings their credibility.

Kargaev (2026) assembled a triaged evidence corpus divided into two layers. The core quantitative layer retained only studies that reported quantitative data on ranking or visibility factors, analysed at least 1,000 queries or URLs, and provided enough methodological transparency to support normalisation. Four studies met this standard: the Aggarwal et al. (2024) GEO benchmark from KDD ’24, Ahrefs’ AI brand visibility study across 75,000 brands, Backlinko’s 11.8 million search result analysis, and the Semrush 2024 ranking factors study. A contextual layer added market context from sources including seoClarity (2025), BrightEdge (2025a; 2025b), Authoritas (2025), and SparkToro (2026).

The paper then maps all retained studies to a four-part factor taxonomy: authority signals (backlinks, domain authority, brand entity mentions, brand search volume), content signals (content quality, content length, in-content citations, in-content statistics, freshness), technical signals (page speed, mobile optimisation, HTTPS, schema markup), and engagement signals (CTR, dwell time).

To compare across studies that use incompatible metrics, the paper introduces a Normalised Importance Score (NIS): each factor’s value within a study is divided by the maximum value in that study, putting all signals on a 0–1 scale. These are then aggregated into an Aggregated Paradigm Score (APS) for each paradigm and compared through the Divergence Index:

DI = APS_GEO − APS_SEO

Values above +0.3 are classified as GEO-ascending. Values below −0.3 are GEO-declining. Values in between are broadly persistent across paradigms.

The paper is explicit about its limitations: this is exploratory, not definitive. The evidence base is still small. Some factors can only be assigned directional interpretations rather than a computed DI. But it is the most rigorous cross-paradigm comparison currently available — and it is far more disciplined than the practitioner commentary most “is SEO dead” discussions rely on.

The Organic Foundation Effect: SEO Is Still the Infrastructure

The single most important finding for anyone wondering whether SEO is dead is what Kargaev (2026) calls the organic foundation effect, drawing on overlap research from seoClarity (2025).

seoClarity’s analysis of Google AI Overviews found that AI-generated answers overwhelmingly include at least one URL that already performs well in Google’s organic results. The generative interface does not reach past the indexed, organically-visible web to find sources. It starts there. GEO, on the current evidence, is not a separate system running independently of organic search — it is a selection and presentation layer built on top of the organic infrastructure that SEO creates.

The practical implication is stark: if your website is not indexed, trusted, and visible in organic search, it is not in the candidate pool from which AI systems draw when generating answers. Skipping SEO to focus exclusively on GEO tactics is not a strategic shortcut — it is building on a foundation that does not exist.

This finding helps explain why authority signals show a Divergence Index of +0.136 — categorised as broadly persistent across paradigms. Kargaev’s composite authority measure, drawing from Backlinko’s domain rating data and Semrush’s domain authority score, produces an APS of 0.261 for SEO and 0.397 for GEO. Authority has not collapsed in the AI era. It has persisted, and in the GEO context may be slightly more salient than classic SEO studies captured — but it remains recognisably the same family of signals.

For SEO practitioners, this is both validating and clarifying. The Backlinko corpus — analysing 11.8 million Google search results — still yields backlinks as NIS 1.000 and referring domains as NIS 0.871 on the SEO side (Backlinko, 2020). These are the strongest link-family signals in the retained corpus. The question is not whether they matter — they clearly do for organic rankings — but whether their mechanism of influence extends directly into the GEO layer or operates indirectly through the organic foundation effect.

The paper’s answer is indirect but real. Kargaev (2026) argues that link signals shape organic discoverability and domain prominence, and those in turn influence the candidate pool from which generative systems draw. What weakens is the explanatory power of backlinks as a direct surface-level GEO signal. Backlinks do not disappear from the picture; they recede one step in the causal chain.

What Has Actually Changed: The Reweighting

If SEO signals persist, what is actually new? The answer from the research is specific and striking.

Brand Entity Mentions: The Dominant GEO Signal

The most dramatic finding in Kargaev’s synthesis comes from the Ahrefs AI brand visibility study, which analysed 75,000 brands across ChatGPT, AI Mode, and AI Overviews. In that study, Brand Entity Mentions — the presence and frequency of a brand being referenced across the broader web, independent of formal link relationships — scores a normalised importance score of 0.918 on the GEO side. This is the highest extracted GEO signal in the entire corpus (Ahrefs, 2025, as cited in Kargaev, 2026).

By comparison, Domain Rating — the classic link-based authority proxy — scores only 0.397 in the same study. Brand Search Volume scores 0.547. The gap between brand mentions (0.918) and domain authority (0.397) on the GEO side is one of the clearest quantitative indicators that the nature of authority is changing.

Kargaev (2026) offers a cautious but important interpretation: AI systems may reward broad web presence and entity recognisability more directly than traditional organic ranking studies do. A business that has accumulated significant editorial mentions across respected publications — not necessarily all with formal backlinks — appears to enjoy substantially higher AI search visibility than a business of comparable domain authority that has not built that distributed brand presence.

This has two practical implications that the paper identifies. First, GEO may be less favourable to publishing strategies that depend on ranking isolated pages without a broader brand footprint. Second, optimisation may need to shift from narrow page-level tactics toward distributed evidence of legitimacy across the web. Authoritas (2025) points in a similar direction, emphasising expert and entity-linked signals in AI search visibility.

In-Content Statistics and Citations: The New Content Signals

The Aggarwal et al. (2024) GEO benchmark, published at KDD ’24 and drawing on 10,000 queries across nine generative AI systems, provides the most direct experimental evidence of what content modifications improve AI search visibility. The paper’s intervention findings, as synthesised by Kargaev (2026), are revealing.

Statistics Addition — adding quantitative data and empirical claims to content — produces a normalised importance score of 0.747. It is the strongest single content intervention in the GEO benchmark. Cite Sources — adding formal references and citations to content — scores 0.671. Fluency Optimization — improving the overall quality and readability of writing — scores 0.684.

Kargaev (2026) interprets these findings with appropriate care: generative retrieval and selection processes do not merely reward relevant content. They reward content that looks quotable, supportable, and evidence-bearing. The suggestion is that AI systems have made legible a set of preferences that always existed implicitly in search quality systems — trust, expertise, usefulness — but were less directly measurable in traditional SEO contexts.

This interpretation is reinforced by research in natural language processing. Gao et al. (2023) demonstrated that citation-capable generation in large language models can be significantly improved through explicit system design, suggesting that the citation behaviour of generative AI is not accidental but reflects deliberate training signals toward attributable, evidence-grounded content. A page that provides a citable statistic with a clear source is more useful to a generative system than a page that makes the same claim without attribution.

The practical implication is direct: content strategy for AI search visibility should prioritise evidence-bearing content — specific data, attributed claims, embedded references — over content that is merely comprehensive or long.

What Has Not Changed Much: Technical Signals

One of the more counterintuitive findings in Kargaev’s synthesis is the near-null contribution of technical factors to GEO visibility. HTTPS scores NIS 0.015 on the SEO side (Semrush, 2024, as cited in Kargaev, 2026) — already low as a ranking differentiator in traditional SEO. Content length scores NIS 0.043. Page speed, in the Backlinko corpus, scores NIS 0.000 within the first-page distribution.

Kargaev’s interpretation is that technical and low-level on-page hygiene factors remain weak differentiators once broader relevance and authority families are considered. HTTPS and page speed are baseline requirements — they need to be correct, but being correct provides no competitive advantage. The businesses spending significant optimisation energy on technical minutiae while neglecting brand entity signals and citation-ready content structure are, on this evidence, misallocating their efforts.

This does not mean technical SEO is irrelevant. The organic foundation effect means that crawlability, indexation, and technical accessibility remain prerequisites for GEO visibility. But the marginal return on technical optimisation — beyond meeting baseline standards — appears low in both SEO and GEO contexts.

The Five Myths This Research Debunks

The evidence base assembled by Kargaev (2026) allows for something unusual: specific, research-grounded rebuttals of specific popular claims. Here are the five most common ones.

Myth 1: “Backlinks No Longer Matter in AI Search”

This is the most widely circulated claim in the “SEO is dead” camp — and the research does not support it. Backlinks remain the strongest signal in the SEO evidence base, with NIS 1.000 in the Backlinko corpus. Their influence on GEO visibility is indirect but real: they build the organic prominence that determines whether a domain is in the candidate pool generative systems draw from. The paper’s conclusion is that backlinks matter more indirectly, not that they have stopped mattering.

Myth 2: “Long-Form Content Is a GEO Signal”

Content length scores NIS 0.043 in the Semrush corpus — effectively negligible. The research strongly suggests that content length as a standalone signal is not what AI systems are selecting for. What matters is content quality and evidential density — the presence of citable statistics, embedded references, and fluent expert explanations. A well-cited 1,500-word piece may substantially outperform an uncited 4,000-word piece in AI search visibility.

Myth 3: “Technical SEO Is Irrelevant in the AI Era”

Technical SEO remains the prerequisite infrastructure that makes both organic rankings and GEO visibility possible. The organic foundation effect documented by seoClarity (2025) means that pages which are not indexed, accessible, and technically sound are not in the candidate pool from which AI systems draw. Technical SEO has shifted from competitive differentiator to table-stakes requirement — but removing it from the strategy is not an option.

Myth 4: “Brand Mentions Replace Domain Authority”

The research does not support a replacement relationship — it suggests an evolution. Domain authority remains a meaningful signal in GEO (DI +0.136, classified as persistent). Brand entity mentions emerge as a distinctly stronger GEO-side signal. The two appear to be complementary: domain authority contributes to organic foundation visibility, while brand entity signals contribute more directly to AI citation and recommendation likelihood. Optimising for one at the expense of the other is a false choice.

Myth 5: “You Can Skip SEO and Go Straight to GEO”

The organic foundation effect makes this impossible on the current evidence. GEO visibility depends on being in the indexed, organically-visible candidate pool. Businesses without functional SEO foundations — poor crawlability, thin content, low domain authority — cannot expect GEO tactics to compensate. The correct model is sequential: build SEO foundations first, then layer GEO-specific signals on top. SparkToro (2026) adds a further caution: citation exposure in AI systems is highly volatile for lower-authority domains, suggesting that weak organic foundations produce unstable AI visibility even when some GEO signals are present.

Separating Two Jobs That Used to Be One

Perhaps the most practically useful framing in Kargaev (2026) is the distinction between two jobs that traditional SEO blended together and that now need to be managed separately.

The first job is ranking eligibility: being discovered, crawled, indexed, and competitive enough in organic search to enter the candidate pool that generative systems draw from. This is still governed by the traditional SEO factor family — technical accessibility, content relevance, domain authority, backlink profile. Nothing in the research suggests this job has become less important. If anything, the organic foundation effect makes it more fundamental, because it is the precondition for everything else.

The second job is citation eligibility: being structured, attributed, and authoritative enough that a generative system selects your content to quote, summarise, or recommend. This is governed by the new GEO-specific signals — brand entity mentions, in-content statistics, embedded citations, fluent expert writing. This is the job that most businesses have not yet started.

The businesses that are winning in AI search visibility in 2026 are doing both jobs well. Those that are struggling are typically doing one job adequately and ignoring the other.

The research makes clear which businesses are at most risk from the SEO-to-GEO transition. As Kargaev (2026) notes in his discussion of brand and entity visibility, GEO may be less favourable to publishing strategies that depend on ranking isolated pages without a broader brand footprint. A business built entirely around keyword-targeted content, without a coherent brand identity, distributed external mentions, or citation-ready content structure, is building ranking eligibility while neglecting citation eligibility. That gap will widen as generative interfaces become more prevalent.

What This Means for Your Strategy in 2026

Translating the research into practical priorities produces a clear hierarchy.

Keep doing — and possibly intensify:

• Building domain authority through editorially earned backlinks. The organic foundation effect means these still determine whether you are in the candidate pool.
• Content relevance and topical authority. Semrush’s Text Relevance remains the strongest SEO-side content signal (NIS 1.000), and content quality is a meaningful GEO signal (NIS 0.684).
• Technical SEO fundamentals. Not for competitive differentiation, but as the prerequisite for indexation and crawlability that makes everything else possible.

Add — these are the new priorities:

• Brand entity signals. Ensure your business is consistently and accurately represented across the web — structured data, directories, editorial mentions, Google Business Profile, knowledge graph signals. Brand Entity Mentions is the strongest measured GEO signal (NIS 0.918).
• Citations and statistics in content. Add quantitative data with clear sources. Add references to authoritative external studies. Statistics Addition (NIS 0.747) and Cite Sources (NIS 0.671) are among the highest-impact GEO content interventions measured.
• Digital PR for distributed brand presence. Earning editorial mentions in publications that AI systems treat as authoritative sources builds both the brand entity signals and the organic authority that GEO requires.

Stop prioritising:

• Content length as a standalone metric. NIS 0.043 does not justify word count inflation at the expense of evidential quality.
• HTTPS optimisation beyond basic compliance. NIS 0.015 — a baseline requirement, not a differentiator.
• Page speed optimisation beyond Core Web Vitals thresholds. Near-null within-page distribution in the Backlinko corpus.

The Verdict From the Evidence: A Direct Answer

So — is SEO dead? Let us be direct.

No. But the question itself contains a false premise that makes it harder to answer usefully. When people ask whether SEO is dead, they are usually asking one of three related but distinct questions: Has Google’s organic ranking algorithm stopped working? Have backlinks and domain authority lost their influence? And is it still worth investing in traditional search optimization when AI platforms are capturing a growing share of discovery?

The research answers all three.

Google’s organic ranking algorithm has not stopped working. BrightEdge (2025a) found that Google search usage increased 49% following the rollout of AI Overviews — the single most cited piece of evidence for the “SEO is dead” narrative is accompanied by data showing Google search volume growing, not declining. The platform is more used than ever, even as the experience of using it changes.

Backlinks and domain authority have not lost their influence — they have shifted how they exert it. On the SEO side, backlinks (NIS 1.000) and referring domains (NIS 0.871) in the Backlinko corpus remain the strongest extracted signals in the retained evidence base. The Divergence Index for the authority family is +0.136 — persistent across paradigms. What has changed is that backlinks now operate partly through the organic foundation effect: they establish the domain prominence that makes a site eligible for AI citation rather than directly causing AI citation themselves.

Is it still worth investing in SEO when AI search is growing? Yes — for exactly the reason the organic foundation effect explains. AI systems draw from the indexed, organically-visible web. A business without SEO foundations has no AI visibility ceiling to reach. The question is not SEO or AI — the question is whether your business is building ranking eligibility and citation eligibility in parallel.

This is what makes the reweighting model so much more useful than either the “SEO is dead” narrative or the “nothing has changed” narrative. Reweighting tells you what to keep, what to add, and what to stop over-investing in. Replacement tells you to abandon the foundation. Continuity tells you nothing needs to change. Only reweighting gives you a usable strategy.

How AIO Clicks Applies This Research

Who Is AIO Clicks?

AIO Clicks is a premium digital visibility agency headquartered in Haaksbergen, Netherlands, serving businesses across the EU — from the Benelux and DACH regions to France, the UK, Scandinavia, and beyond. Founded by entrepreneurs who had operated active B2B and B2C businesses themselves, AIO Clicks was built to solve the problem that the research above describes in quantitative terms: most businesses have the SEO foundation but lack the citation eligibility layer. Most are doing one job and not the other.

The methodology AIO Clicks uses is grounded in exactly the evidence base this post draws on. The finding that brand entity mentions are the strongest measured GEO signal (NIS 0.918) is why Brand Entity Optimization is a core component of the AI Search & GEO service. The finding that in-content statistics and citations are the strongest GEO content interventions is why AIO Clicks’ content strategy prioritises evidence-bearing, citation-ready content over generic long-form production. The organic foundation effect is why Google Rankings & SEO is never treated as optional — it is the infrastructure layer that makes everything else possible.

AIO Clicks works with a focused client base across the EU — going deep rather than wide. Every client works directly with the specialists who built the methodology. No account managers, no generic playbooks. The goal, always, is the outcome the research points toward: businesses that are both ranking-eligible and citation-eligible in a search landscape that increasingly rewards both.

AIO Clicks Services

Google Rankings & SEO — the organic foundation layer. Technical SEO, content architecture, keyword strategy, link building and digital PR, on-page optimisation, and local SEO. Everything that determines whether your domain is in the candidate pool that generative systems draw from.

AI Search & GEO — the citation eligibility layer. Generative Engine Optimization, Answer Engine Optimization, Brand Entity Optimization, schema and structured data implementation, Google AI Overview optimisation, and AI visibility monitoring. Everything that determines whether your content is selected, cited, and recommended by AI systems.

Run the free scan at aioclicks.com/free-analysis to find out whether your business currently has ranking eligibility, citation eligibility, both, or neither — personalised results in 60 seconds, no software required.

Frequently Asked Questions

Conclusion: Reweighting Is Not the Same as Dying

The evidence assembled by Kargaev (2026) — drawing on the GEO benchmark from Aggarwal et al. (2024), Ahrefs’ 75,000-brand AI visibility study, Backlinko’s 11.8 million search result corpus, Semrush’s ranking factors research, and contextual data from seoClarity, BrightEdge, SparkToro, and Authoritas — does not support the conclusion that SEO is dead. It supports a more precise and more actionable conclusion: the factor families that matter are being reweighted, and businesses that understand the reweighting can adapt to it.

Authority persists, with a Divergence Index of +0.136. The organic foundation effect means SEO infrastructure remains the prerequisite for AI search visibility. Brand entity signals emerge as the dominant new GEO factor. Evidence-bearing content — statistics, citations, attributable claims — becomes a distinct citation eligibility signal. Technical signals retreat to baseline-requirement status.

The businesses that treat SEO and GEO as competing priorities — believing they must choose between them — are misreading the research. The businesses that treat SEO as the infrastructure and GEO as the citation eligibility layer built on top of it are aligned with what the evidence actually shows. That is where the competitive advantage lies in 2026: not in abandoning what worked, but in adding what the new search landscape requires.

Find out where your business stands across both layers. Run the free scan at aioclicks.com/free-analysis — personalised results covering traditional SEO health and AI search visibility, in 60 seconds.

Audit Your Website – Free

References

Aggarwal, P., Maatouk, A., Maillard, Q., Gagnon, L., Pal, C., & Boussioux, L. (2024). GEO: Generative engine optimization. Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD ’24). https://doi.org/10.1145/3637528.3671900

Ahrefs. (2025). Top brand visibility factors in ChatGPT, AI Mode, and AI Overviews. https://ahrefs.com/blog/ai-brand-visibility-correlations/

Authoritas. (2025). Can you fake expertise in AI search? We tested 9 models to find out. https://www.authoritas.com/blog/can-you-fake-it-til-you-make-it-in-the-age-of-ai-search

Backlinko. (2020). We analyzed 11.8 million Google search results. https://backlinko.com/search-engine-ranking

BrightEdge. (2025a). One year into Google AI Overviews, BrightEdge data reveals Google search usage increases by 49%. https://www.brightedge.com/news/press-releases/one-year-google-ai-overviews-brightedge-data-reveals-google-search

BrightEdge. (2025b). AI search visits surging in 2025, but organic search remains the cornerstone of digital growth. https://www.brightedge.com/resources/research-reports/ai-search-visits-in-surging-2025

Gao, T., Yen, H. W., Yu, J., & Chen, D. (2023). Enabling large language models to generate text with citations. Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing (EMNLP 2023). https://doi.org/10.18653/v1/2023.emnlp-main.398

Kargaev, D. (2026). The SEO-to-GEO gap: Quantifying ranking factor divergence between traditional and generative search. SSRN. https://doi.org/10.2139/ssrn.6476021

Reyes-Lillo, D., Morales-Vargas, A., & Rovira, C. (2023). Reliability of domain authority scores calculated by Moz, Semrush, and Ahrefs. El Profesional de la Información. https://doi.org/10.3145/epi.2023.jul.03

Semrush. (2024). Ranking factors study 2024. https://seventy2digital.com/wp-content/uploads/2024/01/2024-Google-Ranking-Factors-Study-By-Semrush-English.pdf

seoClarity. (2025). Impact of Google’s AI Overviews: SEO research study. https://www.seoclarity.net/research/ai-overviews-impact

SparkToro. (2026). AIs are highly inconsistent when recommending brands or products; marketers should take care when tracking AI visibility. https://sparktoro.com/blog/new-research-ais-are-highly-inconsistent-when-recommending-brands-or-products-marketers

Published by AIO Clicks — Digital Visibility Specialists | Haaksbergen, Netherlands | aioclicks.com