SEO in 2026: from Google's PageRank (1998) to E-E-A-T and AI Overviews, what still works and what has changed

SEO —Search Engine Optimization— is the discipline of making a website appear prominently in search results for queries relevant to the target audience. The definition hasn't changed in nearly three decades, but what constitutes effective SEO has transformed dramatically with each generation of Google algorithms and, more recently, with the arrival of generative AI search.

This guide covers the industrial history of SEO (because it still affects current decisions), the durable principles that have survived all the algorithmic changes, the key concepts that usually aren't explained well (E-E-A-T, Core Web Vitals, search intent), and the reality of SEO in 2026 with generative AI transforming how information is discovered.

A brief but useful history of SEO

Knowing how we got to 2026 helps explain why certain practices do work and others —which seem logical— don't.

1990s: the pre-Google era. The first search engines (AltaVista, Lycos, Excite, Yahoo!) used relatively simple algorithms based on word frequency and meta tags. It was trivial to manipulate the ranking with keyword stuffing.

September 1998: Google is born. Larry Page and Sergey Brin published "The Anatomy of a Large-Scale Hypertextual Web Search Engine" in April 1998, describing PageRank. They patented the algorithm in 1998. The innovation: using the web's link structure (each link is a "vote") for ranking, not just the content of the page. The quality of Google's results was orders of magnitude better than the competition's. The company grew exponentially.

2000s: classic SEO. The era where building backlinks (quality or quantity), using the right keywords in title tags and the H1, and having a basic semantic structure was enough to rank. Problematic practices grew: link farms, link exchanges, subtle keyword stuffing, content written for the algorithm rather than for humans.

February 2011: Google Panda. A major update that penalized low-quality content, "content farm" sites that produced massive numbers of articles with no substance. Sites like eHow and Demand Media lost significantly. It changed incentives: producing lots of poor content stopped working.

April 2012: Google Penguin. It attacked manipulative link building. Sites with purchased backlinks, link farms, and over-optimized anchor text were penalized. It forced the SEO industry toward more qualified link building.

September 2013: Hummingbird. A significant rewrite of the core algorithm. Better semantic understanding of queries. It marked the start of Google understanding search intent beyond literal keywords.

October 2015: RankBrain. A machine-learning component incorporated into the core algorithm. Google itself acknowledged it was the third most important ranking factor. It improved the interpretation of novel queries (~15% of daily queries that Google had never seen before).

2016-2018: Mobile-first indexing. Google began using the mobile version of sites as the basis for ranking. Sites without a decent mobile version suffered significantly.

October 2019: BERT. A natural-language processing model (Bidirectional Encoder Representations from Transformers). It improved the comprehension of queries with prepositions and contextual nuances.

2020: Core Web Vitals announced. User-experience metrics (LCP, FID, CLS) that would be incorporated as a ranking factor. They forced the industry to take performance seriously. Gradual rollout during 2021.

May 2021: MUM (Multitask Unified Model). A more advanced, multilingual multitask model. Google announced it is 1,000 times more powerful than BERT. Gradual application.

August 2022: Helpful Content Update. A direct penalty for content written for SEO instead of for users. A particularly significant update because it attacked the dominant "SEO content farm" model.

December 2022: E-E-A-T adds the first 'E'. Google's Search Quality Rater Guidelines added Experience to the previous E-A-T formula (Expertise, Authoritativeness, Trustworthiness). The framework became E-E-A-T, emphasizing personal/direct experience with the topic as a quality factor.

March and September 2024: Core Updates focused on spam and low-quality AI-generated content. Google began penalizing content written en masse by AI with no added value. Some sites lost 90% of their traffic overnight.

May 2024: AI Overviews launched in Google Search. AI-generated summaries above the organic results for many queries. It transformed what "ranking" means — a site can be cited in the AI Overview but receive fewer clicks because the answer is given on the SERP itself.

2025-2026: SEO in the era of generative AI. ChatGPT, Perplexity, Claude, Gemini, and others have created new discovery channels. The question "how do I appear in AI answers?" has emerged alongside "how do I appear in Google?". The term GEO (Generative Engine Optimization) has started to be used for this new dimension.

What still works (the durable principles)

Despite all the changes, there are SEO principles that have been valid for 25 years and remain so in 2026:

Genuinely satisfying user intent. Still the number-one factor. A page that truly answers what the user was looking for beats one that is technically perfect but doesn't answer the question. Covered in detail in how long it takes a blog to rank with SEO.

Content quality over quantity. A lesson from Panda 2011 reinforced by Helpful Content 2022. One deep article that truly helps beats 50 generic articles.

Clear semantic structure. An H1 with the main question, H2/H3 that organize sections, lists where they apply, readable paragraphs. It helps both Google and the user.

Quality backlinks. Although Penguin attacked manipulation, links remain an important factor. Editorial backlinks (someone links because your content is good) are the winning pattern.

Speed and mobile experience. Core Web Vitals are the formalization, but the idea —that slow sites perform worse— has been true since 2010.

Smart internal linking. Building topical clusters (cornerstone + spokes) helps Google understand the site's topical authority. Covered in long tail.

Well-built metadata. Title tags attractive for CTR (not just keywords), meta descriptions that promise real value, schema markup where it applies.

Freshness when it applies. Some topics require updating (technology, regulation, data). Evergreen content doesn't need a refresh; time-sensitive content does.

Topical coherence of the site. Sites that cover a niche in depth perform better than sites that touch many topics superficially. Topical authority is an increasingly important signal.

What has changed (what no longer works)

Equally important: what stopped working and many professionals apply out of inertia:

Keyword stuffing. Repeating the keyword every few sentences. Penalized since Panda. Today it hurts performance more than it helps.

Exact match domains (EMD) as an advantage. Having the domain "buycheapshoes.com" doesn't give an advantage on its own, not since the 2012 EMD update.

Purchased or low-quality backlinks. Penguin destroyed the practice. In 2026, disavow tools and detection sophistication make buying links a net negative in most cases.

Generic AI-generated content without curation. Especially since the 2024 core updates. Producing massive numbers of articles with ChatGPT with no differential value is counterproductive.

Article spinning / recycled content. Taking existing articles and rewriting them mechanically. Easily detected, penalized.

Thin, disconnected pages. Creating hundreds of pages to "rank long-tail" without real depth. Helpful Content penalized it.

Pure optimization for keyword density. Density is no longer a significant signal.

Cloaking (showing different content to Google than to users). Sophisticated detection, manual penalties by Google's spam teams.

Doorway pages (pages created exclusively to capture traffic that redirects). Explicitly penalized.

The E-E-A-T framework in detail

Google's Quality Rater Guidelines (published officially for the first time in 2015, updated regularly, with the last significant update in December 2022) introduced and expanded the E-E-A-T framework, which has become a reference for evaluating content quality. The four letters:

Experience (added in 2022). Does the author have direct experience with the topic? A restaurant review written by someone who ate there carries more weight than one synthesized from other reviews. A programming guide written by a developer who has applied it in projects carries more weight than an academic one. The additional E was Google recognizing that first-person experience is a valuable signal.

Expertise. Does the author have technical or professional knowledge of the topic? For YMYL topics (Your Money or Your Life — health, finance, legal matters), expertise becomes critical. A medical article must be written or reviewed by a healthcare professional.

Authoritativeness. Is the source recognized as a reference in its niche? This includes inbound links from authoritative sites, mentions in respected media, presence in relevant professional directories.

Trustworthiness. Is the site trustworthy? Honest information, transparency about sources, a clear privacy policy, verifiable contact info, HTTPS, the absence of malware or deceptive practices. T is considered the most fundamental — without trust, none of the others matter.

For a brand that produces content:

• Show real authors with relevant bios and credentials.
• Cite sources verifiably.
• Keep information up to date.
• Have a readable privacy policy and terms.
• Offer clear contact channels.
• Build a presence on relevant professional platforms (LinkedIn for B2B, industry certifications, etc.).
• Document direct experience when it applies (your own cases, primary data, field observations).

Core Web Vitals: the technical dimension

Announced by Google in May 2020 and incorporated as a ranking factor during 2021-2022, Core Web Vitals measure user experience in three dimensions:

Largest Contentful Paint (LCP). Time until the largest content element is visible. Good: ≤ 2.5 seconds. Needs improvement: 2.5-4 seconds. Poor: > 4 seconds.

Interaction to Next Paint (INP). It replaced First Input Delay (FID) in March 2024. It measures how quickly the page responds to user interactions. Good: ≤ 200ms. Needs improvement: 200-500ms. Poor: > 500ms.

Cumulative Layout Shift (CLS). It measures visual stability during loading. Pages where elements jump while loading are frustrating. Good: ≤ 0.1. Needs improvement: 0.1-0.25. Poor: > 0.25.

These metrics are measured with real user data (CrUX — Chrome User Experience Report) or with tools (PageSpeed Insights, Lighthouse, WebPageTest, GTmetrix). The actual factor in ranking is modest, but the improvement typically correlates with better engagement, which does affect it indirectly.

The most impactful levers for improving Core Web Vitals:

• Image optimization (a modern format like WebP/AVIF, lazy loading, appropriate dimensions).
• Aggressive caching and a CDN.
• Reduction of blocking JavaScript.
• Careful management of embeds and third-party scripts (covered in embed).
• Design that avoids layout shifts (reserving space for images, ads, etc.).
• A reasonable server response time.

Search intent: the most important factor

Any serious introduction to SEO in 2026 must emphasize search intent over literal keywords. As TOFU covered, Google classifies search intents into four main categories:

Informational. The user is looking for information. "What is X", "how does Y work". The majority of queries.

Navigational. The user is trying to reach a specific site. "HubSpot login", "Polimake pricing".

Transactional. The user wants to make a transaction. "Buy X", "hire Y".

Commercial. Research with commercial intent. "Best X 2026", "X review", "X vs Y".

Before optimizing for a keyword, you have to verify what kind of results Google currently shows for that query. If the top 10 are long informational articles, Google has decided the intent is informational. Optimizing a sales page for that keyword won't work no matter how much technical SEO you apply.

Researching the SERP (Search Engine Results Page) before producing content is a basic step. Tools like Ahrefs, Semrush, and ContentKing have SERP analysis that speeds up this work, but you can also do it simply by searching manually.

Current SEO tools

The tools that professionals typically use in 2026:

Search Console (free, Google). Essential. It shows how your site appears in Google, which queries bring traffic, what technical errors there are, what backlinks you have.

Google Analytics 4 (free). Covered in Google Analytics. For analyzing the behavior of users who arrive via SEO.

Ahrefs. A complete suite for competitor research, backlink analysis, keyword discovery, technical audits. Typical plan from ~€100/month.

Semrush. A direct competitor of Ahrefs with similar functionality. Similar pricing.

Moz Pro. The oldest of the trio, still a valid option. Historically strong in local SEO.

Screaming Frog SEO Spider. A local crawler for detailed technical audits. Free for small sites, ~€250/year for large ones.

Sitebulb, OnCrawl, Lumar (formerly DeepCrawl). Technical SEO platforms for large sites.

Surfer SEO, Clearscope, MarketMuse. Content-optimization tools based on SERP analysis.

Google Trends (free). Covered in Google Trends. For validating interest and seasonality.

PageSpeed Insights (free, Google). For measuring Core Web Vitals.

Schema.org markup validators. For verifying structured data.

For a small business, Search Console + GA4 + the free version of Screaming Frog is usually enough for basic SEO. For mid-sized companies, Ahrefs or Semrush add significant capability. For large companies, combinations of several platforms.

The 2026 reality: SEO in the era of generative AI

Something that has changed dramatically since 2023 is the entry of generative AI into the search ecosystem:

ChatGPT, Perplexity, Claude, Gemini as sources of information. A growing portion of users look for information by asking these chatbots instead of Google. The answers often cite sources (Perplexity especially), but the click-through to the source is typically lower than in traditional Google.

Google's AI Overviews (May 2024). For many queries, Google shows an AI-generated answer above the results. If your content is cited in the AI Overview, there's less chance the user clicks through to your site (the answer is already on the SERP).

The term GEO (Generative Engine Optimization) has emerged to describe the discipline of optimizing to appear in and be well represented in AI answers. The practices:

• Content structured so that the AI can cite specific segments.
• Factual information clearly documented.
• Rich schema markup that facilitates interpretation.
• Building recognizable topical authority.
• Presence in sources that AIs typically use (Wikipedia, authoritative niche sites).

The impact on organic traffic. Data from several studies suggests that AI Overviews are reducing the CTR of organic results by 20-50% for affected queries. Sites whose business model depended 100% on organic traffic are reevaluating their strategy.

The new challenge: how does a content business sustain itself when a growing proportion of searches generate no click? The answers are still evolving — more emphasis on direct brand, on owned email lists, on monetization that doesn't depend on traffic, on content so unique that users seek out the source.

Common SEO mistakes in 2026

Doing SEO without a content strategy. Optimizing pages technically without having decided what you want to communicate.

Producing massive AI content without human curation. Especially penalized since 2024. Generic AI-generated content doesn't rank.

Ignoring Core Web Vitals because "content is what matters". Content is what matters, but content on a slow site gets read less and performs worse.

Obsessing over keywords instead of intent. Optimizing for "X review" when the SERP shows that the real intent is a comparison, a listicle, or buying.

Not updating old content. Posts that ranked two years ago may be losing positions due to being outdated. A refresh usually pays off more than creating new content.

Purchased or low-quality backlinks. It still happens; it still doesn't work and remains risky.

Nonexistent schema markup. Structured data helps Google understand content. Its absence is an opportunity cost.

Not optimizing images. Camera photos uploaded without resizing kill Core Web Vitals.

Ignoring internal linking. Every isolated article with no internal links to related pieces is an underused asset.

Doing only on-page SEO without building topical authority. The site that covers a topic with 10 deep interconnected articles performs better than the one that covers 100 topics superficially.

Not measuring beyond total traffic. Traffic is an input, not a result. Conversions, qualified leads, and contribution to the business are the metrics that matter.

Confusing SEO with marketing. SEO is a channel. Marketing is a discipline. Optimizing SEO without a broader strategy rarely produces a business result.

Assuming SEO is "set and forget". Algorithms change, competition improves, queries evolve. SEO requires continuous maintenance.

Not diversifying discovery channels. A company that depends 100% on Google is fragile. SEO + email + social + paid + an owned community is a portfolio.

SEO and creative operations

For a brand that invests in SEO, the regular production of quality content is the engine. Without that engine, technical SEO alone doesn't produce sustained results. And producing quality content at scale requires operational infrastructure.

That coordination is a discipline of creative operations: the editorial calendar coordinates the consistent production that SEO rewards, content production sustains a regular flow with quality, and creative KPIs measure whether the content produces a business result, not just vanity traffic.

At Polimake, that logic lives across three surfaces: Studio coordinates SEO production with other priorities, Studio produces pieces with a consistent brand system, and Media stores reusable material that reduces the marginal cost of production.

---

If you lead marketing, content, or digital strategy and you got here looking for an answer about SEO, the most useful thing you can take from this article is probably the combination of three ideas: effective SEO in 2026 is indistinguishable from honest content strategy (answering real intents well, with demonstrated experience and authority, on technically solid sites), the fundamental principles have been valid for two decades (quality over quantity, intent over keywords, authority over tricks), and the significant changes come from generative AI (AI Overviews and conversational chatbots) that are redefining what "appearing in search" means. SEO as a discipline exists; SEO as a trick-the-search-engine-in-your-favor works less and less.

To round it out, how long it takes a blog to rank with SEO covers the time reality of SEO work, Google Trends covers the complementary trends tool, and long tail covers the specific keyword strategy that has proven to work.

Quick references

• How long it takes a blog to rank with SEO — the time reality of SEO.
• Google Trends — a complementary trends tool.
• Long tail — a specific keyword strategy that works.
• Exit rate — a behavior metric that indirectly affects SEO.
• Google Analytics — the complementary measurement tool.
• Google Business Profile — local SEO specifically.