On-Page SEO
Title tags, meta descriptions, heading hierarchy and content optimization for AI extraction.
For: Content managers, marketing teams, SEO beginners
On-page SEO optimizes all directly controllable elements on a webpage, from title tags to content structure to internal linking.
Key Takeaways
- ✓Google rewrites about 76% of title tags (Q1 2025)
- ✓Organic CTR drops by 61% for queries with AI Overviews
- ✓Question-based headings improve AI extraction
- ✓Entity-based SEO replaces keyword density
- ✓Optimal passage length for AI Overviews: 134–167 words
On-Page SEO: The Enterprise Guide to Content Optimization, Structure, and AI Visibility
On-page SEO encompasses all optimization measures that take place directly on your own website, from keyword strategy to content structure to technical page elements like title tags, meta descriptions, and schema markup. It is the area of search engine optimization over which you have complete control. No waiting for backlinks, no dependence on external factors, just systematic optimization of what you can directly influence.
But on-page SEO has fundamentally changed since 2024. Google rewrites 76% of all title tags. 60-70% of meta descriptions are auto-generated. The Helpful Content System has been folded into core ranking: Google mainly evaluates individual pages and also draws on site-wide signals. And AI search systems cite pages with correct heading hierarchy 2.8x more frequently than pages without clear structure. Anyone still doing on-page SEO like it's 2020 is optimizing the wrong levers.
This guide covers all aspects of modern on-page SEO, backed by current data, algorithm leaks, and empirical studies. Each section links to in-depth articles with concrete implementation guides.
Keywords: From Strings to Entities
Keyword optimization has fundamentally transformed. Google no longer understands search queries as isolated character strings but as semantic concepts (entities). The search engine recognizes that "artificial intelligence," "AI," "KI," and "machine intelligence" refer to the same concept, and evaluates pages by how well they cover that concept, not how often a specific keyword appears.
In practice, this means keyword stuffing doesn't help and can even hurt. Instead, it's about semantic coverage: comprehensively treating a topic including related terms, subtopics, and contextual connections. Topic modeling, TF-IDF analysis, and entity mapping are the tools of modern keyword strategy. How keywords have evolved from strings to entities and what this means for your content strategy is covered in Keywords: From Strings to Entities and Semantic Understanding.
Title Tags: Understanding Google's 76% Rewrite Rate
The title tag remains one of the strongest on-page ranking factors. Yet an analysis of Q1 2025 data published on Search Engine Land shows: Google rewrites 76% of all title tags in search results. This means the title you define in HTML is not the title users see in SERPs in three out of four cases.
Google generates alternative titles from various sources: H1 headings, anchor texts of incoming links, Open Graph tags, and page content itself. The rewrite occurs when Google considers its own title better. The consequence: title tags must be consistent with the H1, directly address search intent, and stay within the optimal length of 50-60 characters. Google's rewriting logic is explained in Title Tags 2026: Google Rewrites 76%, How to Prevent It.
Meta Descriptions: Still Relevant Despite Auto-Generation
Meta descriptions aren't a direct ranking factor, but they influence click-through rate (CTR) in search results, which in turn influences rankings as a user signal. Studies show Google replaces 60-70% of all meta descriptions with its own snippets generated from page content.
Optimization still pays off: in the remaining 30-40% of cases, Google shows the original description. A Backlinko analysis of 4 million search results found that pages with a meta description got an average of 5.8% more clicks than pages without one. Details are in Meta Descriptions: 60-70% Are Auto-Generated by Google.
Heading Hierarchy: 2.8x More AI Citations
A page's heading structure (H1-H6) is far more than a design element. It defines the semantic hierarchy of content and signals search engines and AI systems how content is organized. Current data shows: pages with correct, logical heading hierarchy are cited 2.8x more frequently in AI-generated answers.
The rules are simple but surprisingly often violated: one H1 per page, H2 tags for main sections, H3 tags subdividing H2 sections, no skipped levels. Heading texts should clearly name the section's topic, no generic phrases like "Learn more." Why heading hierarchy is so critical for AI citations is explained in Heading Hierarchy: 2.8x More AI Citations with Correct Structure.
URL Structure and Technical On-Page Elements
Beyond visible content elements, technical on-page factors influence rankings: URL structure, canonical tags, meta-robots directives, Open Graph tags, and page architecture. The complete overview is in URL Structure and Technical On-Page Elements.
Internal Linking: The Underestimated On-Page Lever
Internal linking is one of the most underestimated on-page factors. A Zyppy analysis of about 23 million internal links across roughly 1,800 websites found that URLs with 40 to 44 internal links received about four times as many Google clicks as URLs with 0 to 4 links. That is a correlation, not proof of cause. A SearchPilot A/B test on about 8,000 regional pages measured a causal effect: adding links to the geographically nearest regions lifted organic traffic to the tested pages by 7%.
Internal links serve three functions: they distribute link equity from strong pages to weaker ones, help Google with crawling and topical classification, and guide users to relevant further content. The pillar-cluster architecture is the most effective model. Common mistakes include orphan pages without incoming internal links, excessive navigation links instead of in-content links, and generic anchor texts. The full analysis is in our in-depth article on internal linking.
Content Freshness: A Query-Dependent Signal
Google evaluates content freshness as a ranking signal, but how much it matters depends on the query, and Google publishes no fixed weight for it. Freshness doesn't mean changing the publication date without updating content. Google detects substantive changes and distinguishes them from cosmetic updates. The right freshness strategy is covered in our in-depth article on content freshness.
E-E-A-T On-Page: Signaling Trust Directly on the Page
E-E-A-T is often viewed as an off-page topic, but much of it can be implemented directly on-page. The Princeton GEO study (KDD 2024) shows that visible evidence also pays off in AI answers: optimized content was up to 40% more visible in generative engine responses, and citing sources, adding quotations and adding statistics worked best.
On-page E-E-A-T signals include: author bios with credentials, publication dates and "last updated" timestamps, source references, editorial guidelines, and trust signals like contact information and privacy policies. For B2B companies, the Experience aspect is particularly decisive: case studies, proprietary data analyses, and concrete numbers from real projects. How to implement E-E-A-T on-page systematically is covered in our in-depth article on on-page E-E-A-T.
Helpful Content System: Quality at Page and Domain Level
Google introduced the Helpful Content System in August 2022 as a site-wide signal. Since March 2024, it has been part of the core ranking systems. According to Google, these work mainly at the page level and also use site-wide signals. A high share of unhelpful content can therefore still hold back the entire domain. Quality must be comprehensive, not just on the most important pages. What this means for your content strategy is covered in our in-depth article on the Helpful Content System.
User Engagement Signals and NavBoost
The 2024 Google API leak confirmed what the SEO community long suspected: Google uses user interaction data as ranking signals. NavBoost, an internal Google system, evaluates aggregated click data from a 13-month window to determine which search results users prefer, as Google VP Pandu Nayak testified in the US antitrust trial.
If users click your result, stay on the page, and don't return to search results (no pogo-sticking), Google interprets this as a positive signal. The full analysis of NavBoost and its implications is in User Engagement Signals: NavBoost as Confirmed Ranking Factor.
Image and Media Optimization
Images and media are an increasingly important on-page factor, and modern image optimization starts with the format. In a 2020 ctrl.blog comparison of 600 images at equal visual quality, AVIF files were a median 50.3% smaller than JPEG, WebP files 31.5%. By that math, AVIF is roughly 27% smaller than WebP. Add responsive images with srcset attributes, lazy loading, and descriptive alt texts. The complete guide is in Image and Media Optimization: AVIF, Google Lens and Visual Search.
For video SEO, additional on-page requirements apply: VideoObject schema markup, transcripts for accessibility and AI indexing, and optimized video thumbnails. Details are in Video SEO On-Page: VideoObject Schema and Transcripts.
Schema Markup On-Page: Required for Rich Results, Not a Booster for AI Citations
Structured data, preferably in the JSON-LD format Google recommends, translates your content into machine-readable vocabulary and is a prerequisite for rich results. The data does not show that schema increases AI citations. Ahrefs compared 1,885 pages that added JSON-LD with 4,000 control pages: citations in AI Overviews fell by 4.6% relative to the controls, while the changes in AI Mode (+2.4%) and ChatGPT (+2.2%) were statistically indistinguishable from zero. The AI Overviews decline is significant, but both groups were already losing citations before schema was added. For B2B and content sites, the types that pay off most today are Article, BreadcrumbList, Organization and Product, all listed in Google's search gallery. Google shows review snippets only for certain types such as products, software, books or recipes. Reviews a business collects about itself and marks up on its own site get no stars. Google retired HowTo rich results in September 2023 and has stopped showing FAQ rich results altogether since May 7, 2026. Service markup produces no rich result but describes your offering in machine-readable form. Correct implementation decides whether rich results can appear in SERPs. It makes your content easier for AI systems to interpret, but it does not guarantee citations. Details are in our in-depth article on on-page schema markup.
GEO On-Page: The Implementation Framework
Generative Engine Optimization (GEO) is not a separate discipline but an extension of on-page SEO. AI systems prefer content with clear factual statements that can be cited: definition boxes, numbered lists, comparison tables, and paragraphs with clear thesis-evidence structure. The complete framework is in GEO On-Page: The Practical Implementation Framework.
Featured Snippets: Visibility Despite Decline
Featured snippets have lost visibility: according to Ahrefs, their share of US desktop SERPs fell from 15.41% in January to 5.53% in June 2025, a 64% decline, in parallel with the growth of AI Overviews. Where a featured snippet does appear, its CTR is above the position 1 average: 11.4% versus 7.1% according to First Page Sage (as of September 22, 2026). Optimization requires specific content structure: clear Q&A patterns, definition paragraphs of 40-60 words, comparison tables, and numbered lists. When optimization is worthwhile is covered in our in-depth article on featured snippets.
German On-Page Specifics
On-page SEO for the German-speaking market has particularities not found in English best practices: compound words (Komposita), umlauts in URLs and meta tags, and the formal address (Sie vs. Du). The DACH-specific challenges are covered in German On-Page Specifics: Compound Nouns, Umlauts and Formal Address.
Algorithm Updates and On-Page Impact
The Google algorithm updates of 2024-2026 have significantly shifted on-page requirements. A chronological analysis is in Algorithm Updates 2024-2026: The Most Important Changes.
On-Page SEO Auditing: Enterprise Workflow
For enterprise websites with hundreds or thousands of pages, a systematic audit workflow is essential. How to build a scalable on-page audit process is covered in On-Page SEO Auditing: Enterprise Workflow and Continuous Optimization.
Summary: On-Page SEO as Controllable Excellence
On-page SEO is the area of search engine optimization over which you have full control:
- Content & Keywords: Semantic optimization, content freshness as ranking signal, Helpful Content as a page- and domain-level quality standard
- Structure & Markup: Title tags and meta descriptions Google retains, heading hierarchy for AI citability, schema markup for rich results
- Technical Elements: URL structure, internal linking as an underestimated traffic lever, image and media optimization
- E-E-A-T & Engagement: On-page trust signals, NavBoost as confirmed ranking factor, GEO framework for AI visibility
- DACH Specifics: German language particularities, algorithm update analysis, enterprise audit workflows
Start with the highest-impact topics. In most cases, that means title tags, internal linking and E-E-A-T signals.
All Articles in this Topic
18 ArticlesTitle Tags 2026: Google Rewrites 76%, How to Prevent It
A title tag is the HTML element that defines the title of a web page and serves as the clickable headline in search results. Google does not necessarily use this title, though: it rewrites title tags in 76.04 percent of cases and keeps only 35 percent of the original words on average. A rewrite-resistant title tag stays within the 51 to 60 character range, puts the primary keyword first, places the brand name at the end and matches the page's H1.
Meta Descriptions: 60-70% Are Auto-Generated by Google
Meta descriptions are the HTML element that Google can display as a snippet source below the title in search results. They are not a direct ranking factor, but they influence click-through rate. In most cases, Google generates the description it actually displays from the page content itself and only uses the stored meta description when it fits the specific query better.
Heading Hierarchy: 2.8x Higher Citation Odds with Correct Structure, per AirOps
A correct heading hierarchy is the consistent, semantic nesting of headings (exactly one H1, with H2, H3 and so on below it, without skipped levels) that represents the content structure of a page in machine-readable form. It is an SEO signal and today also a central GEO lever: language models use headings as signposts to identify individual passages, extract them and cite them in AI answers.
E-E-A-T On-Page: Sources, Statistics & Quotes for Up to 40% More AI Visibility
E-E-A-T on-page SEO means operationalizing the four quality signals Experience, Expertise, Authoritativeness, and Trust directly in the page template, for example through an author box, source citations, update stamps, trust pages, and schema markup. E-E-A-T is not a single ranking score, but through algorithmic proxies it works as a central ranking influence, and it increasingly decides whether AI Overviews and AI chatbots cite a page as a trustworthy source.
Keywords: From Strings to Entities and Semantic Understanding
Semantic SEO is the optimization of content for meaning and entities instead of character strings and keyword density. Search engines process terms through the Knowledge Graph as linked concepts and assess how central an entity is to a text (entity salience). The goal is to cover a topic completely and be recognized as a distinct entity, instead of placing a keyword as often as possible.
Internal Linking: +7% Organic Traffic in a Controlled A/B Test
Internal linking is the deliberate connection of pages within the same website through crawlable HTML links. It controls which pages search engines discover and crawl and how relevance and link equity (PageRank) are distributed across the site, and it provides topical context through the anchor text. Internal links are by far the most common type of link on the web and a central on-page lever for rankings, crawl control and visibility in AI answers.
Image and Media Optimization: AVIF, Google Lens and Visual Search
Image SEO covers all measures that make images discoverable by search engines, image search, visual search (Google Lens) and multimodal AI systems, and fast to load for users. These include descriptive alt texts (which also serve accessibility), modern image formats such as AVIF with a picture fallback, prioritizing the LCP image for Core Web Vitals, and structured data such as ImageObject and Product schema. Because an image is the LCP element on 76% of mobile pages, image optimization is performance work and visibility work at the same time.
Schema Markup On-Page: Required for Rich Results, Not a Booster for AI Citations
Schema markup on-page refers to structured data that translates a web page's content into a machine-readable vocabulary so that search engines and AI systems understand unambiguously what a page describes. JSON-LD is Google's preferred format and, according to the Web Almanac 2024, was used on 41 percent of pages. Structured data is the technical prerequisite for rich results, but it does not act as a direct booster for AI citations.
Featured Snippets: -64% Visibility, 11.4% CTR When They Appear
Featured snippets are highlighted answer boxes above Google's classic results (often called "position zero") that answer a question directly in search. Their visibility dropped sharply in 2025, from 15.41% to 5.53% of the US SERPs studied, but where they appear, they still reach an 11.4% CTR in 2026 according to First Page Sage: more than position 1 next to an AI answer (3.6%), but only about half as much as position 1 without an AI answer (22.6%). The same extractable on-page structure also serves People Also Ask and AI Overviews.
User Engagement Signals: NavBoost as Confirmed Ranking Factor
User engagement signals are the user behavior measurable on the search results page (clicks and subsequent satisfaction) that Google feeds directly into rankings via its NavBoost system. NavBoost evaluates a rolling 13-month window of aggregated click data and primarily assesses click quality (goodClicks, badClicks, lastLongestClicks) rather than click volume. The system was confirmed by Pandu Nayak's testimony in the DOJ antitrust trial (October 2023) and the Content Warehouse API leak (May 2024).
URL Structure and Technical On-Page Elements
URL structure covers how a web address is built, how long it is and how readable it is, together with the related technical on-page elements such as canonical tags, hreflang, title tags and meta robots. SEO-friendly URLs are short and descriptive and use hyphens as word separators. For the DACH region, self-referencing canonicals and correctly linked hreflang annotations (de-AT, de-DE, de-CH, x-default) prevent duplicate content and language mix-ups.
GEO On-Page: The Practical Implementation Framework
GEO on-page refers to optimizing all on-page elements of a web page (title tags, heading structure, schema markup, internal linking, citation density, and entity coverage) with the goal of both ranking in classic search results and being cited as a source by generative AI systems such as ChatGPT, Perplexity, and Google AI Overviews. It combines established on-page SEO rules with the techniques that the Princeton GEO research found most effective for increasing AI visibility.
Content Freshness On-Page: When Freshness Matters and When It Doesn't
Content freshness describes how current a piece of content is, treated as an on-page relevance signal. Google does not weight freshness the same for every search; it depends on the query (Query Deserves Freshness, QDF, known since 2007 and expanded in 2011 with the Freshness Update): some queries call for fresh results, others do not. What matters is substantial maintenance, not the publication date. Changing only the date without substantive content gives you no ranking advantage, and Google's guidance on helpful content treats it as a warning sign of search engine-first content.
Helpful Content System: Helpfulness in Core Ranking Since March 2024
The Helpful Content System is Google's mechanism for assessing whether content was created primarily for people rather than for search engines. Since March 2024, it is no longer a standalone classifier; it is integrated directly into core ranking. Since then, Google has assessed helpfulness mainly per page and also draws on sitewide signals: a high share of unhelpful content can dampen the visibility of the entire domain. After the update, Google saw 45 percent less low-quality, unoriginal content in search results.
German On-Page Specifics: Compound Nouns, Umlauts and Formal Address
German on-page optimization has to account for linguistic features that English SEO lacks: compound nouns use up title budgets quickly (Suchmaschinenoptimierung, German for search engine optimization, is 24 characters on its own, and German words are significantly longer), umlauts belong in visible content (ä, ö, ü) and are replaced with ae, oe, ue in URLs, and the form of address (formal Sie or informal Du) has to fit the target audience and stay consistent across the entire domain. On top of that, you have to handle regional variants (de-AT vs. de-DE vs. de-CH) and language-specific schema markup with inLanguage de-AT.
Video SEO On-Page: VideoObject Schema and Transcripts
On-page video SEO is the technical and editorial optimization of dedicated watch pages so that search engines and AI crawlers understand a video and surface it in search and generative results. Its core is VideoObject schema with the required fields name, thumbnailUrl and uploadDate, crawlable full-text transcripts, captions and an indexable page structure. Because AI crawlers do not watch videos but read the page text, the transcript is the decisive signal for machine understanding.
On-Page SEO Auditing: Enterprise Workflow and Continuous Optimization
In an enterprise context, on-page SEO auditing is the systematic, ongoing review of all page-level ranking factors (title tags, meta descriptions, heading hierarchy, content structure, internal linking, schema markup, E-E-A-T and GEO signals) across the entire URL inventory. Unlike a one-off audit, it combines a full crawl (e.g., Screaming Frog) with data sources such as Search Console, log files and visibility tools (SISTRIX in the DACH region) and turns the findings into continuous, template-level optimization validated by SEO A/B tests.
Algorithm Updates 2024-2026: The Most Important Changes
Between March 2024 and October 2026, Google rolled out nine core updates, eight spam updates, and one Discover update, according to its Search Status Dashboard; the most recent core update ran from May 21 to June 2, 2026. The March 2024 Core Update had the largest documented effect: it integrated the Helpful Content System into core ranking and introduced three new spam policies (Scaled Content Abuse, Site Reputation Abuse with enforcement from May 5, 2024, and Expired Domain Abuse), after which, according to Google, 45 percent less low-quality content appeared in the results. The Content Warehouse API Leak in May 2024 was not an update, but with 14,014 documented attributes it confirmed that engagement signals (NavBoost) and a stored authority value (siteAuthority) are part of the evaluation. With the rise of AI Overviews, the focus is also shifting from classic ranking to Generative Engine Optimization (GEO), meaning optimization for being cited in AI answers.