How does GEO differ from SEO in 2026?

SEO is the fight for a spot in the top ten blue links. GEO (Generative Engine Optimization) is the fight for a place among the mere 2-7 domains a large language model typically cites in a single AI answer. It is not a replacement of SEO with GEO but a layering — SEO stays the foundation (Perplexity and ChatGPT often pull content from high-ranking pages), and on top of it you build AEO (answer structure) and GEO (AI citability).

Is traffic from AI search worth less than classic organic?

On the contrary. A Semrush study from 9 June 2025 (over 500 topics) found the average AI-search visitor is about 4.4x more valuable by conversion than a classic organic visitor. Seer Interactive reports ChatGPT converting at 15.9% versus 1.76% from Google Organic. AI takes clicks, but the traffic that does reach the site is pre-qualified — the user arrives after the model has already synthesized information and compared options.

What most strongly influences whether AI cites my brand?

Three levers. First, brand mentions in third-party sources: Ahrefs (75,000 brands analyzed) measured a 0.664 correlation between mentions and AI visibility versus only 0.218 for classic backlinks (mentions are ~3x the stronger predictor). Second, density of facts and statistics: the Princeton study (KDD 2024) showed over 40% visibility gain from citing sources. Third, topical authority built through a content cluster (pillar + supporting articles + FAQ).

Do llms.txt and schema.org increase AI citations?

Evidence for their direct impact on AI citations is weak or none. An SE Ranking study (November 2025) found no correlation between llms.txt presence and citations, and Ahrefs (May 2026, 1,885 pages) found that simply adding schema did not meaningfully raise citations in AI Overviews, AI Mode or ChatGPT. Implement both for other benefits (rich results, entities, knowledge graph, disambiguation) but not as a magic switch for AI citations.

What does Google penalize in the AI-content era in 2026?

Google does not penalize AI use. It penalizes scaled content abuse — mass-producing content to manipulate rankings without value to users. John Mueller (November 2025): our systems do not care whether content was made by AI or a human; what matters is whether it is helpful. Every piece must go through human editing, fact-checking and add original value. The March 2026 core update heavily weighted information gain — a signal measuring how much new knowledge content adds relative to what already ranks.

SEO, GEO and AEO in 2026: the complete guide to the new search era and building websites that earn

ARDURA Lab pillar page · Updated: June 2026

TL;DR — the three key takeaways

Search has split. Google still handles most queries (per StatCounter, January 2026 — 90.04% of the global search market across all devices and 79.1% on desktop, the lowest in over 20 years), but AI Overviews and AI search engines (ChatGPT, Gemini, Perplexity) are taking over informational traffic. In Poland AI Overviews appears on about 24% of queries, and organic CTR has dropped roughly 19-35%. Optimizing for "blue links" alone is no longer enough.
GEO/AEO is a new discipline, not a buzzword. Visibility is now measured by citations and mentions in AI answers, not just rankings. The key levers: brand mentions in third-party sources (correlate with AI visibility ~3x more strongly than links), density of facts/statistics in content (+over 40% visibility per the Princeton study) and topical authority.
AI traffic is smaller but more valuable. Visits from AI search convert about 4.4x better than classic organic (Semrush study, 9 June 2025, over 500 topics). The winners combine SEO fundamentals (technical, content, E-E-A-T) with GEO and build sites engineered for conversion (CRO), not just for traffic.

What is GEO and why does it change the rules of the game in 2026?

For over two decades, SEO came down to one thing: ranking as high as possible on a list of ten blue links. In 2025 and 2026 that game changed fundamentally. The user increasingly does not get a list of links — they get a ready, synthesized answer, sometimes inside Google (AI Overviews, AI Mode) and sometimes entirely outside Google (ChatGPT, Perplexity, Gemini, Copilot).

GEO (Generative Engine Optimization) is the practice of optimizing content and brand presence so that AI systems — ChatGPT, Google AI Overviews, Perplexity, Claude, Copilot — choose, cite and recommend your brand when answering user questions. The term was formalized in a paper by a team from Princeton, Georgia Tech and IIT Delhi (Aggarwal et al., "GEO: Generative Engine Optimization", arXiv:2311.09735), presented at the KDD 2024 conference.

AEO (Answer Engine Optimization) is a closely related concept emphasizing appearance in direct answers — featured snippets, "People Also Ask", AI Overviews. In practice the terms GEO, AEO, as well as LLMO or GSO, describe the same phenomenon: the fight for visibility in a world where the answer matters more than the link.

The difference from classic SEO is structural. If SEO is the fight for a place in the top ten, GEO is the fight for a place among the mere 2-7 domains a large language model usually cites in a single answer. Instead of ten chances you have a few. It is a "winner-takes-most" game.

What matters most for a decision-maker: this is not replacing SEO with GEO, but layering them. SEO remains the foundation — if you cannot compete in classic search, AI platforms will not find you either, because Perplexity and ChatGPT often pull content from high-ranking pages. On that foundation you build AEO (answer structure) and GEO (AI citability). More in SEO vs GEO — how they differ and GEO vs SEO — which strategy to choose.

How do AI Overviews and AI Mode affect organic traffic — hard data 2025/2026?

This is the most important change of the last two years and you have to understand it in numbers.

Scale. Google announced at I/O 2025 that AI Overviews reaches 1.5 billion users monthly in over 200 countries. AI Mode — a dedicated, conversational answer interface that runs dozens of sub-queries ("query fan-out") — was rolled out to all US users on 5 June 2025.

CTR drops. A Seer Interactive study (September 2025, over 3,000 informational queries) found that when an AI Overview appears, organic CTR drops from 1.76% to 0.61% — a 61% fall. Ahrefs in December 2025 (analysis of ~300,000 keywords) found AI Overviews reduce CTR for position-1 content by 58%. Other studies confirm the direction: Authoritas (~47.5%), Kevin Indig (>50%).

Zero-click. According to various analyses, about 60% of Google searches end without a click. In AI Mode this reaches as high as 93%.

The Polish perspective. Senuto's report "AI Overviews Analysis in Poland" (nearly 18 million keywords) shows AIO appears on 24.17% of Polish queries, and for simple informational questions (Know Simple) in 57.82% of cases. 19.4% of clicks "evaporated" from the Polish internet (June 2025 vs 2024), and between May and June 2025 alone Polish sites lost 23.7 million organic clicks. 64% of Polish domains (919 of 1,435 analyzed) felt a negative AIO impact on traffic. Importantly for local brands: per Senuto, as much as 96.03% of citations in Polish AI Overviews come from domestic domains, and the average number of citations in a single AIO answer is about 6.86 — a real opportunity for Polish firms that know how to stand out. How to get into those overviews is covered in AI Overviews in Google — how to get featured.

An important nuance. Google claims (Liz Reid, August 2025) that the total volume of clicks from search to sites remains "relatively stable". Publishers disagree — Chegg sued Google, and some publishers report organic traffic drops of tens of percent (HubSpot lost about 70-80% of organic traffic from its peak). The truth is that the impact is asymmetric: informational and educational content loses the most, while commercial and transactional queries are so far less affected.

What it means for business. A position in Google is no longer the same as visibility. You can hold position 1 and lose half your clicks at the same time. That is why KPIs must change: from "position and traffic" to "share of voice in AI, citation frequency and traffic quality".

Is AI search traffic worth less? What does the data say about conversion?

It is a paradox that changes the ROI calculation. Yes, AI takes clicks — but the traffic that does reach the site is significantly more valuable.

A Semrush study (published 9 June 2025, by Kyle Byers and Rachel Handley, over 500 topics) found the average AI-search visitor is 4.4x more valuable (by conversion) than a classic organic visitor. Seer Interactive reports ChatGPT traffic converting at 15.9% versus 1.76% from Google Organic. Ahrefs in its own analysis found that 0.5% of AI-sourced traffic generated 12.1% of sign-ups — about 23x higher conversion.

The mechanism is intuitive: the AI user arrives after the model has already synthesized information from several sources, compared options and "pre-qualified" them. They land on the site later in the funnel and closer to a decision.

A note on methodology. Some studies (e.g. Amsive's analysis of 54 sites, p=0.794) found no significant difference. The AI advantage concentrates in B2B and research-heavy topics and weakens in e-commerce. Semrush forecasts AI traffic may reach parity in value with classic organic around 2027. That is a forecast, not a fact — treat it as a trend, not a certainty.

How to optimize content so AI cites it? Concrete GEO/AEO tactics

This is where the practical part begins. Here is what actually works, grounded in research.

1. Front-loading answers. A Growth Memo analysis (2026) found that 44.2% of all LLM citations come from the first 30% of the text. Answer the question directly in the first 40-60 words of a section, then develop context.

2. Density of facts and statistics. This is the strongest lever from the Princeton study. In Aggarwal et al. (KDD 2024), 9 methods were tested on 10,000 queries (the GEO-bench benchmark), and the conclusion is blunt: "including citations, quotations from relevant sources, and statistics can significantly boost source visibility, with an increase of over 40% across various queries". Citing sources lifted visibility by 30-40%, and adding expert quotations by about 28%. Importantly — keyword stuffing made results worse. Practical rule: one concrete, source-attributed statistic every 150-200 words.

3. Structure for extraction. An SE Ranking study showed that sections 120-180 words long between headings receive far more citations than short fragments, and pages with comparison tables are cited about 2.5x more often. Use a clear H2/H3 hierarchy, lists, tables and FAQ sections with self-contained answers (40-80 words). See our technical SEO checklist.

4. Freshness. Ahrefs (analysis of 17 million citations, July 2025) found content cited by AI is fresher than the organic average, and ChatGPT has the strongest preference for fresh URLs. Update cornerstone content, add new data and an honest "last updated" marker.

5. Entities and unambiguity. Models think in entities and context, not keywords. Build a clear brand identity as an entity (consistent NAP, Organization schema with sameAs, presence in credible sources).

Poland's Senuto added a module for monitoring AI Overviews visibility, and Google consistently repeats that proven SEO methods still apply to AI features and there are no "special" requirements — but there are actions that move you closer to the goal.

Do llms.txt and structured data really help AI visibility?

Here we need to dispel two myths that GEO agencies are happy to sell.

The llms.txt myth. llms.txt is a file meant to point AI models to the most important content. It sounds reasonable, but the data is merciless. An SE Ranking study (November 2025, ~300,000 domains) found only about 10.13% of sites have the file and — crucially — there is no correlation between llms.txt presence and AI citations. Analysis of over 500 million AI bot visits showed that among the user-agents that actually drive citations (GPTBot, ClaudeBot, PerplexityBot, OAI-SearchBot, Google-Extended), a negligible fraction touch the llms.txt file at all. Google (Gary Illyes, July 2025) confirmed it does not support llms.txt, and John Mueller compared it to the long-abandoned keywords meta tag. Verdict: you can implement it (near-zero cost, genuinely optional), but it is not a lever that changes anything today.

The "silver bullet" schema myth. Here it is more nuanced. Ahrefs (May 2026) analyzed 1,885 pages that added JSON-LD and found that simply adding schema did not meaningfully raise citations in AI Overviews, AI Mode or ChatGPT. A searchVIU experiment showed that when pulling a page directly, none of the major AI systems (ChatGPT, Claude, Perplexity, Gemini, Google AI Mode) used hidden schema — they extracted only the visible HTML. Google officially stated that for AI Overviews "you don't need any special schema.org". On the other hand, Bing confirmed (March 2025) that schema helps its LLMs (Copilot), and Google admits structured data "gives an edge in results".

Practical verdict: structured data still makes sense — for rich results, voice assistants, knowledge graph and entity disambiguation. Implement Article, Organization, Product and FAQPage where they fit. But treat schema as hygiene and a trust/entity signal, not a magic switch for AI citations. Update: after the March 2026 core update Google narrowed rich-results eligibility (including FAQ and How-To), and on 7 May 2026 officially retired FAQ rich results. Full picture in our schema.org guide.

Technical SEO 2026: what must work so AI bots and Google can even see you?

The foundation has not changed — only the range of bots you must serve has.

Crawlability for AI bots. Check in robots.txt whether you are blocking the user-agents that drive citations: GPTBot and OAI-SearchBot (OpenAI), ClaudeBot and Claude-User (Anthropic), PerplexityBot, Google-Extended, Applebot-Extended. This is the most common and most costly mistake — blocking these bots means invisibility in AI. Decide consciously whether to allow training (Disallow: GPTBot) separately from on-demand citation (Allow: ChatGPT-User). Details in our technical SEO checklist.

Rendering. Most AI bots do not execute JavaScript as well as Googlebot. Critical content should be in server-side rendering or prerendered. If content exists only client-side, for many models it simply does not exist.

Core Web Vitals and INP. Since March 2024, INP (Interaction to Next Paint) replaced FID as the responsiveness metric. "Good" thresholds: LCP < 2.5 s, INP < 200 ms, CLS < 0.1. This is not just ranking — it is conversion. According to case studies, improving CWV can raise conversion by low double-digit percentages (Renault: ~13% conversion lift after improving LCP; redBus: +7% sales after optimizing INP), and Google reports that sites meeting the thresholds have a 24% lower bounce rate. Only a minority of sites meet all three thresholds (per analyses ~46-48%), creating a real advantage for those that do. How to improve it — in our Core Web Vitals guide.

Indexing and IndexNow. Because ChatGPT Search uses the Bing index, it is worth implementing the IndexNow protocol to speed up indexing in Bing — without it, local and fresh content may be unavailable for ChatGPT queries.

E-E-A-T, core updates and AI content: what Google penalizes and rewards in 2026

The core updates calendar. After 2024 Google released: March 2025, June 2025, December 2025, and in 2026 — the first ever Discover-only update (February), March 2026 and the May 2026 core update (completed 2 June 2026). March 2026 was the most volatile core update in history — per SE Ranking data (100,000 keywords) 79.5% of top-3 URLs changed position, and 24.1% of top-10 pages fell out of the top 100.

E-E-A-T and information gain. The March 2026 update heavily weighted "information gain" — a signal measuring how much new knowledge content adds relative to what already ranks. Pages that merely repackage existing information lose. Winners: content with named authors of verifiable expertise, based on first-person experience and proprietary data. In September 2025 Google expanded YMYL to include "Government, Civics & Society". Of the four E-E-A-T pillars, Trust is the most important — which Google confirms outright in its "people-first content" documentation (update of 10 December 2025).

AI content — what does Google actually penalize? This is where the biggest misunderstanding lives. Google does not penalize AI use. It penalizes "scaled content abuse" — mass-producing content to manipulate rankings without value to users. John Mueller (November 2025): "Our systems do not care whether content was created by AI or a human. What matters is whether it is helpful." Google's official documentation states outright that using generative AI to create many pages without added value may violate the scaled content abuse policy (sections 4.6.5 and 4.6.6 of the rater guidelines). June 2025 brought documented manual actions for "large-scale content abuse".

Takeaway for companies: AI as a tool for research and drafting — yes. AI as a factory for thousands of templated pages — a straight road to a penalty. Every piece must pass through human editing, fact-checking and add original value. Worth remembering: research (Ahrefs, 600,000 pages) shows no correlation between AI use and lower rankings — quality matters, not the production method.

Why have brand mentions become more important than links in the AI era?

This is probably the biggest change in off-page SEO since PageRank.

Ahrefs analyzed 75,000 brands and measured what best predicts presence in AI answers. The result: brand mentions across the web correlate with AI visibility at 0.664, while classic backlinks at only 0.218 — meaning mentions are about 3x the stronger predictor. In a December 2025 update, the single strongest signal turned out to be YouTube mentions (~0.737). The number of subpages on a site has near-zero correlation — content volume does not build AI visibility.

The mechanism: models learn from raw text, not from a link graph. When independent sources consistently describe a brand in a given context, the model "learns" that it is a credible entity worth recommending. Patrick Stox of Ahrefs called this "the era of off-page SEO", and Tim Soulo (CMO of Ahrefs) — "citations are the new links".

The key consequence: most brand mentions in AI answers come from third-party sources, not from the brand's own site. The AirOps study (Oshen Davidson, 17 October 2025, 21,311 mentions across over 500 commercial queries on GPT-5, Claude Sonnet 4.5 and Perplexity Sonar) reports: "85% of brand mentions came from external domains, while only 13.2% of mentions came directly from the brand's domain". Independent sources confirm it (Muck Rack: 82% of citations from earned media; Omniscient Digital: 91% from third parties). This reverses two decades of SEO strategy. Your content on your own domain matters for classic search, but brand trust in AI is built by independent sources: media, industry portals, reviews (G2, Capterra, Trustpilot), Reddit, YouTube, Wikipedia.

Since mentions beat links, the off-page budget must shift from mass link acquisition to earning authentic mentions in credible sources.

Where AI searches most. AI citations concentrate on a narrow set of authority domains. Per a Semrush study (June 2025, citation analysis in ChatGPT, Perplexity, Gemini and AI Overviews) the most-cited domains are Reddit (about 40% of answers containing a citation), Wikipedia (about 26%) and YouTube (about 24%) — though shares are unstable and differ between engines. Reddit is, per Profound data, the most-cited domain in Google AI Overviews and Perplexity and the second in ChatGPT — related in part to the Google-Reddit licensing deal (February 2024, about USD 60M per year for access to Reddit content for AI products). SE Ranking (November 2025, ~129,000 domains) found that "domains with profiles on platforms like Trustpilot, G2, Capterra, Sitejabber, and Yelp have 3x higher chances of being chosen by ChatGPT as a source".

Proof it works. A controlled Stacker/Scrunch study (16 March 2026, 87 stories, 30 brands, over 2,600 prompts, 8 AI platforms) showed that distributing content through earned-media channels delivers a median 239% lift in AI search visibility. 97% of distributed stories earned at least one AI citation versus 82% for content on the brand's own domain (p < 0.006).

What to do in practice:

Citation-focused digital PR: expert commentary, proprietary data, studies and reports that journalists and creators happily cite with attribution. This builds links, mentions and entity context at once.
YouTube: mentions in reviews, comparisons and tutorials — models read transcripts, and YouTube is the single strongest signal.
Reddit and Quora: authentic participation, not spam — moderation and models quickly filter promotional content.
Review platforms: a well-kept G2/Capterra/Trustpilot profile is a third-party validation signal, especially important in B2B and SaaS.
Wikipedia: if the brand meets notability criteria, a Wikipedia page strongly reinforces entity identity.

A timing nuance: brand-level mentions take time — often 3-6 months before models "notice" new social content, and 4-12 months before the change is visible in ChatGPT, Perplexity and AI Overviews. It is a long-term investment.

Topical authority and topic clusters: why does depth beat single articles?

In 2026 the most durable advantage in search is not another link or a perfect title tag — it is becoming so comprehensively authoritative on a topic that both Google and AI systems cite you by default.

The March 2026 core update made E-E-A-T and topical depth key factors. The rule Google enforces: a site with 20 related articles on a topic consistently outranks a site with one, even excellent, 5,000-word guide. AI networks behave analogously — when ChatGPT or Perplexity need a source on a topic, they reach for sites recognized as consistently accurate and comprehensive.

The pillar-cluster (hub-and-spoke) model:

Pillar page (like this article) — broadly covers a large topic at a strategic level.
Cluster articles — deepen individual subtopics and link back to the pillar.
Internal linking — binds the cluster, signaling to Google which content matters most.

This is exactly the approach ARDURA Lab uses in SaaS and B2B projects: pillar page + supporting articles + case studies + comparisons + FAQ. Such a cluster signals expertise to Google and AI models, and after a few months becomes a lead-generation machine. In our experience, clusters that cross the topical-authority threshold after 6-9 months start getting cited by AI without any extra link budget — the moment the strategy begins to pay for itself.

How to measure AI visibility? New KPIs and LLM tracking tools

Old KPIs (position, traffic, CTR) are not enough, because Google Search Console does not distinguish clicks from organic results versus AI Overviews. You need two parallel measurement tracks.

Track 1 — classic: positions, organic traffic, conversions (GSC, GA4, Senuto, Ahrefs, Semrush).

Track 2 — AI visibility:

Citation frequency — how often AI cites/mentions the brand.
Share of voice in AI — your share of answers versus competitors.
Citation sources — which pages AI cites on your topic.
AI referral traffic — in GA4 create a channel group filtering chat.openai.com, perplexity.ai, gemini.google.com, copilot.microsoft.com, claude.ai.
Indirect signals — growth in branded queries and direct traffic.

Tools. The market has matured: Profound (enterprise class, from about USD 99/mo, full coverage of 10+ platforms in higher plans), Peec AI (mid-market, from about EUR 89-100/mo), Otterly.AI (entry-level, from about USD 29/mo, 6 platforms). Ahrefs Brand Radar tracks mentions across several AI indexes. In Poland, Senuto lets you check on how many phrases your domains trigger an AI Overview.

Important context: citations are unstable — per various analyses 40-60% of cited domains change month to month. That is why the trend matters, not a single measurement. One mention is a curiosity; systematic growth is proof the strategy works.

Local SEO and Google Business Profile in the AI era: what changes for companies?

Local search remains a strong Google stronghold. About 46% of all searches have local intent (figures given by a Google representative at the "Secrets of Local Search" conference in 2018, cited by Search Engine Roundtable), and for queries like "plumber near me" Google still prefers a fast Local Pack over a generative answer — per some analyses, AI Overviews triggers on only about 7% of direct local queries.

The layer beneath has changed, though. Google Business Profile (GBP) data now feeds Gemini-based AI Overviews answers. A well-optimized profile not only wins a Local Pack spot — it supplies structured facts to the AI answer layer. More in local SEO and Google Business Profile.

Local priorities 2026:

Complete GBP: categories (the key primary category), description, services with descriptions, hours, photos (Google uses Vision AI for verification), regular posts.
Reviews: quantity, freshness and authenticity — AI analyzes sentiment and keywords in opinions. Note: Google tightened enforcement in 2026 against keyword stuffing in business names.
NAP consistency and LocalBusiness schema aligned with GBP data — inconsistencies lower the "entity trust score".
Hyperlocal content answering real customer questions (location pages must be unique, not templated).

Building websites that earn: how to combine SEO, GEO and CRO?

Visibility without conversion is vanity. In a world where traffic is scarcer but more valuable, every visitor must count. That is why strategy must combine three layers.

1. Technical foundation (Crawl). Speed, INP, clean code, correct indexing, accessibility for AI bots. Without it neither Google nor AI models will see the content.

2. Content and authority (Trust). Topical authority, E-E-A-T, information gain, citability. This attracts traffic from both channels — classic and AI.

3. Conversion (Convert). Here we measure revenue, not positions. Mastery of INP is not an SEO task but a CRO function — every millisecond shaved off interaction is a direct investment in the financial result. Information architecture, clear CTAs, landing pages designed for a specific intent, forms and conversion paths.

Measuring ROI from SEO/GEO. The most mature approach (and the one ARDURA Lab uses in B2B projects) is combining SEO data with the CRM: tracking the path from an organic entry, through trial/lead, to a closed customer, and reporting not positions but MQL/SQL counts and customer acquisition cost (CAC) from organic versus paid. That is a conversation the board understands. In our experience on B2B projects, companies that report leads and CAC instead of positions are far less likely to cut budget in a panic after an AI-Overviews-driven session drop — because they can see conversions and traffic quality holding firm.

The future of search: AI vs Google market share and forecasts

Where we are. Google still dominates (StatCounter, January 2026: 90.04% of the global search market across all devices, 79.1% on desktop — the lowest in over two decades). But for the first time in over two decades its hegemony has been genuinely challenged: per the First Page Sage report from Q4 2025, ChatGPT accounts for about 17% of all digital queries globally. ChatGPT has about 900 million weekly active users (official OpenAI statement, February 2026 via Reuters — a doubling from 400 million in February 2025) and processed about 2.5 billion prompts daily (OpenAI's last official figure, July 2025).

The AI market is fragmenting. Per Similarweb, ChatGPT's share of generative traffic fell from about 87% (early 2025) to about 64.5% (early 2026), while Gemini grew more than threefold (from about 5.7% to about 21.5%). It is no longer one player's monopoly but a competitive ecosystem (ChatGPT, Gemini, Perplexity, Claude, Copilot, Grok).

Poland. ChatGPT exploded: from 3.6 million real users in January 2025 to 9.3 million in June 2025 (about 31% of internet users), and in October 2025 it passed 10 million (34.4% reach) — Gemius/PBI Mediapanel data. Notably, 39.6% of users live in rural areas, and users aged 25-34 spend over 2.5 hours per month with the tool. It is a daily-use tool, not a curiosity.

Forecasts (with caution). Gartner predicted (statement of 19 February 2024, Alan Antin, VP Analyst): "By 2026, traditional search engine volume will drop 25%, with search marketing losing market share to AI chatbots and other virtual agents". Semrush forecasts AI traffic may surpass classic organic around 2028. These are forecasts in the conditional mood — treat them as scenarios, not certainties. What is certain is the direction: search is fragmenting, and the discipline splits into classic (transactional, navigational — Google's domain) and conversational/informational (AI's growing domain).

How ARDURA Lab helps you win in the new search era (SEO + GEO)

Most SEO agencies still report positions and traffic. That is not enough when you can be #1 in Google and lose half your clicks to an AI Overview. ARDURA Lab approaches it differently — we combine classic SEO with GEO and CRO, focused on one thing: measurable impact on revenue.

What we do:

SEO and GEO audit. We diagnose technical fundamentals (crawlability for Googlebot and AI bots, INP, indexing), content quality (E-E-A-T, information gain, citability) and visibility in AI search. You get a prioritized list of concrete actions, not generalities.
Content strategy and topical authority. We build topic clusters (pillar + supporting articles + case studies + comparisons + FAQ) designed to rank in Google and be cited by AI.
Technical optimization. We make sure both Google and GPTBot, ClaudeBot or PerplexityBot can correctly read and interpret your site.
Building sites that convert. We design information architecture, landing pages and conversion paths aimed at leads and sales, not just traffic.
Digital PR for AI citability. We help you earn mentions and presence in the sources AI actually cites.
Business reporting. We combine SEO data with the CRM — showing MQLs, SQLs and CAC from organic, not just positions.

See our GEO service and digital strategy.

Next step: order a free SEO/GEO audit of your site. We will show where you are losing visibility (in Google and in AI) and propose a strategy tailored to your B2B business. In an era where search changes every quarter, inaction is not a strategy — and the longer you delay adapting, the harder the losses are to recover.

Recommendations: a step-by-step action plan

Stage 1 (0-30 days) — diagnosis and fundamentals.

Run a technical audit: check robots.txt for AI bots (GPTBot, ClaudeBot, PerplexityBot, OAI-SearchBot, Google-Extended), measure INP/LCP/CLS, fix rendering of critical content (SSR).
Set up AI visibility measurement: a GA4 channel group for AI domains + one tracking tool (start with Otterly or Senuto for the Polish market).
Decision threshold: if CWV is below thresholds or AI bots are blocked — that is priority #1, before you invest in anything else.

Stage 2 (1-3 months) — content and structure.

Redesign key content for citability: front-loaded answers, sourced statistics every 150-200 words, 120-180-word sections, comparison tables, FAQ.
Plan a topic cluster around the pillar page; begin systematic publishing.
Decision threshold: if after 90 days content is not appearing in AI citations and positions are not rising — verify information gain (whether you add something competitors lack).

Stage 3 (3-12 months) — authority and off-page.

Launch digital PR for mentions: proprietary data, expert commentary, presence on G2/Capterra/Trustpilot, YouTube, Reddit (authentically), Wikipedia (if you qualify).
Monitor share of voice in AI versus competitors and adjust.
Decision threshold: if organic traffic falls but conversions and leads rise — that signals the strategy is working (the "great decoupling"); do not cut budget based on a session drop alone.

What would change these recommendations: if your business is purely local/service-based (e.g. a workshop, a clinic), shift the weight to GBP and reviews instead of content GEO. If you are e-commerce, prioritize product schema, speed and classic transactional SEO — the AI conversion advantage is weaker there than in B2B/SaaS.

Caveats — methodological reservations

AI impact data is fresh and unstable. Citation patterns change dramatically (e.g. after September 2025 — removal of the num=100 parameter and a change to OpenAI's retriever). Treat any domain shares in citations as a snapshot, not a permanent state. 40-60% of cited domains change month to month.
Two different citation-measurement methodologies. "% of answers containing a domain" (Semrush: Reddit ~40%) is different from "% of all citation events" (Profound/Goodie: Reddit/Wikipedia 2-8%). Both are valid but not directly comparable.
Conflict of sources on AI Overviews' traffic impact. Google (Liz Reid) claims clicks are "stable"; independent studies (Seer, Ahrefs, Senuto) show CTR drops of 47-61%. We lean toward the independent data, but the difference stems partly from different metrics (total volume vs CTR on AIO queries).
Much data comes from GEO/AEO tool vendors (Profound, Otterly, SE Ranking, AirOps, Stacker) who have an interest in promoting the topic. The strongest findings are those where independent methods agree (e.g. 82-91% of mentions from earned media per AirOps, Muck Rack and Omniscient).
AI conversion — conflicting results. Semrush (4.4x) and Ahrefs (23x) vs Amsive (no significant difference, p=0.794). The advantage is real in B2B/research, weaker in e-commerce.
Forecasts are forecasts. Gartner's figures (25% drop by 2026) and Semrush's (crossover around 2027-2028) use the conditional mood and are not accomplished facts. Some traffic-decline forecasts are already being questioned (total volume is not falling as assumed — click distribution is shifting).
Schema and llms.txt: evidence for their direct impact on AI citations is weak or none (Ahrefs, SE Ranking). Implement them for other benefits (rich results, entities), not as a GEO lever.