61% of the titles you publish today will be rewritten by Google before the user sees them in search results. That is not anecdotal data: it is a direct signal that on-page SEO is being done badly on a massive scale. If Google needs to rewrite your title, it means it isn’t fulfilling its function. And if the most visible element of your page doesn’t work, the rest of the on-page optimizations won’t perform at their best either.
This guide covers the three on-page elements with the greatest direct impact on rankings and CTR: title tags, meta descriptions and header structure. With data, formulas and real examples, not generic theory.
The title tag: the on-page element with the highest ROI
The title tag is the only piece of content that works simultaneously on two fronts: it acts as a relevance signal for Google’s algorithm and as the first visual hook for the user scanning results. No other on-page element has that dual impact.
The SISTRIX study of more than 80 million keywords shows that position 1 in pure organic results achieves a CTR of 28.5%, while position 2 drops to 15.7% and position 3 to 11%. That difference of 13 percentage points between first and second position is largely due to title quality: a well-constructed title can outperform a competitor holding a higher position in CTR.
Optimal length: the character vs pixel trap
The “60 characters” rule circulating in most guides is imprecise. Google measures available space in pixels, not characters. A title with many wide letters (M, W) takes up more space than one with narrow letters (i, l). For English, the practical reference is:
- Safe zone: 50-58 characters
- Desktop limit: ~600px wide (equivalent to about 55-60 characters in standard font)
- Mobile limit: more restrictive, ~40-50 visible characters
The key data point: Google truncates titles that exceed the available space with ”…”, which can make the most important keyword disappear if it’s at the end.
Title formulas that work in English
The two structures with the best documented performance:
Formula 1: Number + Benefit + Keyword
“7 link building techniques that work in 2026” “5 technical SEO errors blocking your indexation”
Formula 2: Keyword: Promise
“Title tags: how to prevent Google from rewriting them” “SEO audit: what to review before publishing”
The first formula works especially well for list content where the number creates specific expectation. The second is more effective for informational content where the promise resolves a specific problem.
Good vs bad title tag examples
| Title (bad) | Problem | Title (good) | Why it works |
|---|---|---|---|
| ”SEO Services” | Generic, no benefit or specific keyword | ”Technical SEO for ecommerce: audit from £490” | Keyword + specificity + price |
| ”Welcome to our digital marketing blog” | Useless boilerplate, homepage title in category | ”Digital marketing in Barcelona: 2026 guide” | Local + relevance + action |
| ”How to optimize your site for Google and Bing and Yahoo and other internet search engines” | Keyword stuffing, exceeds 80 characters | ”How to optimize your site for search engines: practical guide” | Concise, clear intent |
| ”Page 1” | No information for user or Google | ”Off-page SEO strategies: 8 proven tactics” | Number + specificity |
How to prevent Google from rewriting your title
Google officially documents the rewrite triggers: empty or semi-empty titles, titles with outdated text, titles that don’t match the page content, boilerplate repeated across multiple pages, and titles in a different language from the content. The most effective way to prevent rewriting is to align the title exactly with the search intent of the target keyword. If your title describes what the page actually contains, Google rarely has reason to change it.
Meta descriptions that generate real clicks
The meta description is not technical SEO: it’s direct response copywriting. Google confirmed years ago that it doesn’t use it as a ranking factor. However, it’s the second most visible element in SERPs after the title, and its unique function is to convince the user to click on your result instead of a competitor’s.
That is the mental reframing you need: the meta description is a free text ad in the most competitive space on the internet. And like any ad, it can be optimized.
Proven structure: problem → solution → implicit CTA
The structure that consistently generates the highest CTR has three components:
- Recognizable problem: Identify the pain or question the user has in mind
- Specific solution: Show that your page contains the concrete, not generic, answer
- Implicit CTA: A call to action that doesn’t sound forced (“Discover…,” “Learn…,” “Check the data…”)
Example for the keyword “technical SEO audit”:
Bad: “On this page you’ll find information about technical SEO audits and how to perform them step by step to improve your website.” Good: “Your site is losing visibility through invisible technical errors. Discover the 12 critical points of a technical SEO audit, with tools and real examples.”
The difference: the second identifies the problem (invisible errors), specifies the content (12 points, tools, examples) and creates implicit urgency.
150-160 characters: how to count
The practical rule is to count including spaces and every character as one. For greater safety, aim for 150-155 characters to leave margin for rendering variations.
Quick tool: paste your meta description into the Google search bar (without submitting) to preview how it will appear truncated. The article on what is technical SEO places the meta description within the broader context of optimisation.
The meta description rewrite problem
Google rewrites meta descriptions far more frequently than titles, in up to 70% of searches according to some studies, selecting page content fragments it considers most relevant for the specific query. This means that even if your meta description is perfectly written, Google may ignore it. The solution isn’t to give up writing meta descriptions, but to ensure the article body also contains well-structured passages that Google can select when it decides not to use yours.
The H1: primary signal of thematic relevance
If the title talks to Google (and to the user in SERPs), the H1 talks to the user within the page. They are distinct elements with distinct purposes, though they are often confused.
H1 vs title tag: why they can and should be different
The title is optimized for CTR on the results page: it must be concise, capture attention amid the competition and fit the available space. The H1 is optimized for the in-page experience: it can be longer, more descriptive and more oriented toward the content the user will find after clicking.
Practical example:
- Title (for SERPs): “Link building in 2026: 8 strategies that work”
- H1 (for the page): “The 8 most effective link building strategies for English-language sites in 2026”
The H1 can add context, be more specific, or reframe the angle for the reader who has already clicked.
The single H1 rule
A page must have exactly one H1. Multiple H1s on the same page send confusing signals to Google about what the main topic is. It’s a frequent error in WordPress themes where widget or module titles are incorrectly marked as H1.
For the H1, the optimal length is between 40-70 characters. Less than 40 can be too vague; more than 70 can dilute the thematic focus. The primary keyword should appear in the H1, but it doesn’t need to be an exact match: Google understands semantic variations and synonyms.
Semantic variations in the H1
For a target keyword like “on-page SEO optimization,” the H1 doesn’t need to repeat that exact string. Valid variations that Google recognizes as equivalent:
- “How to optimize the on-page SEO of your website”
- “On-page optimization guide: title, meta and headers”
- “On-page SEO: the factors that matter most in 2026”
All three contain the core concept without keyword stuffing and are semantically equivalent for the algorithm.
H2-H3 header structure: semantic architecture
Headers are not typographic decoration: they are the structure Google uses to understand the content architecture of your page. A well-structured article with coherent H2s and H3s functions like an index that tells Google what subtopics it covers and how they relate to each other.
How headers define semantics for Google
Google uses headers to build its understanding of a page’s thematic coverage. If your article on “SEO audit” has H2s covering crawling, indexation, speed, security and structured data, it’s signaling to Google that it addresses the topic comprehensively. That improves relevance for related queries beyond the main keyword.
The hierarchy rule is strict: never skip levels. H1 → H2 → H3 → H4, in that order. An H4 directly under an H2 (skipping H3) indicates a fragmented semantic structure. In practice, most blog articles don’t need to go beyond H3.
Question-format headers for featured snippets and AI Overviews
Question-format headers (How…? What is…? When…?) have a specific advantage: they increase the probability of capturing Google featured snippets and being cited in AI Overviews.
The optimal structure is: H2/H3 with question → direct answer in the first paragraph of that section. Google extracts the question-answer pair for the snippet or AI synthesis. If the first paragraph under the header doesn’t contain the answer directly, Google will look for another candidate.
The article on how to appear in Google AI Overviews goes deeper on optimisation for AI-generated responses.
Recommended number of H2s per article
There is no prescribed number, but there is a useful reference: in a 2,000-word article, between 4 and 7 H2 sections results in sections of 280-500 words, which is the range where Google can extract fragments with sufficient context. Sections of less than 150 words can be too superficial; sections of more than 700 words can lose thematic focus.
On-page + GEO: optimizing for AI citation
GEO (Generative Engine Optimization) doesn’t replace traditional on-page SEO: it extends it. The language models powering Google AI Overviews, ChatGPT Search and Perplexity operate on the same content that Google crawls and indexes. But they apply different criteria for selecting what to cite.
How AI models choose what to cite
AI citation studies identify three characteristics that increase the probability of a content fragment being extracted and cited:
1. Verifiable data with sources: Claims with specific numbers and explicit source (source: X) are cited more frequently because models assign them higher reliability.
2. Self-contained passages: A paragraph that makes complete sense read in isolation, without needing the context of previous paragraphs, is extractable without distortion.
3. Direct answers in the first paragraph of each section: According to Semrush, content with verifiable quantitative data is 47% more likely to appear in AI Overviews.
Technique: passage-level optimization
Passage-level optimization means writing each section with the mindset that it will be read in isolation. Concrete structure:
- First sentence of the paragraph: main claim (the answer)
- Second-third sentence: data or evidence supporting it
- Last sentence: practical implication or context
This structure is easily extractable by AI models and simultaneously improves readability for human users.
Schema.org FAQ as a citability booster
FAQ schema markup not only enables rich results in Google: it also helps language models identify the explicitly marked question-answer structure in the HTML. A well-marked FAQ pair with schema has a higher probability of being cited in AI Overviews than the same pair without markup. The guide on how to implement schema markup covers the JSON-LD technical details for the most relevant types.
On-page checklist of the 10 most important points
Effective on-page optimization doesn’t require reviewing 50 elements: it requires correctly executing the 10 that have the most impact. For each point, a verification tool.
1. Unique and descriptive title tag (50-58 characters) The primary keyword should appear in the first 60 characters. No duplicates on other pages of the site. Tool: Screaming Frog → Title filter / Google Search Console → Coverage report
2. Single H1 with primary keyword (40-70 characters) One H1 instance per page. Semantic variation of the main keyword. Tool: Screaming Frog → H1 report / SEOquake extension
3. Meta description with problem-solution-CTA structure (150-160 characters) Unique per page. Don’t repeat the title verbatim. Tool: Screaming Frog → Meta Description / Google Search Console
4. Primary keyword in the first paragraph (first 100 words) Doesn’t need to be the first sentence, but must appear naturally in the introduction. Tool: manual Ctrl+F search on the published page
5. Primary keyword in the URL (slug) Short URL, lowercase, no special characters, hyphens instead of spaces. Tool: manual review / Screaming Frog
6. Primary keyword in the alt text of the main image The hero/featured image alt text should describe the image and include the keyword naturally. Tool: Screaming Frog → Images / element inspection in browser
7. H2-H3 header structure without skipping levels H1 → H2 → H3 without interruptions. Each H2 covers a distinct subtopic. Tool: HeadingsMap extension for Chrome/Firefox
8. Internal linking to related pages (minimum 2-3 links) Relevant context, descriptive anchor text, not generic (“click here”). Tool: Screaming Frog → Inlinks / manual review
9. Structured data (Article + FAQ when applicable) Schema.org Article for blog posts. FAQPage if the content includes questions and answers. Tool: Google Rich Results Test
10. Load speed: LCP < 2.5s A page with good on-page SEO but LCP > 4s is penalized in rankings due to Core Web Vitals. Tool: Google PageSpeed Insights / CrUX in Search Console
Where to place the primary keyword: visual summary
| Element | Keyword? | Note |
|---|---|---|
| Title tag | Yes, ideally at the start | First 30 characters if possible |
| H1 | Yes, semantic variation | Exact match not required |
| URL (slug) | Yes, short form | Without unnecessary stopwords |
| First paragraph | Yes, first 100 words | Natural, not forced |
| Main image alt | Yes, descriptive | Describes the image, not just the keyword |
| Meta description | Yes, preferably | Doesn’t affect ranking, does affect CTR |
| Main H2 | Yes, at least one | Can be in LSI variation |
| Body text | Yes, natural density | No stuffing (2-3% maximum) |
The three on-page elements with the highest ROI per hour of work are, in order: the title tag, the H1, and the header structure. SEO does not produce identical results across every sector or market. What works for an ecommerce store may not work for a law firm. If you want to know which factors carry the most weight in your specific case, tell us about your situation.
Share this article
If you found this content useful, share it with your colleagues.
Frequently Asked Questions
¿Con qué frecuencia publican contenido nuevo?
Publicamos artículos nuevos semanalmente, enfocados en las últimas tendencias de SEO técnico, casos de estudio reales y mejores prácticas. Suscríbete a nuestro newsletter para no perderte ninguna actualización.
¿Los consejos son aplicables a cualquier tipo de sitio web?
Nuestros consejos se adaptan a diferentes tipos de sitios: ecommerce, blogs, sitios corporativos y aplicaciones web. Siempre indicamos cuándo una técnica es específica para cierto tipo de sitio o requerimientos técnicos.
¿Puedo implementar estas técnicas yo mismo?
Muchas técnicas básicas puedes implementarlas tú mismo siguiendo nuestras guías paso a paso. Para optimizaciones avanzadas o auditorías completas, recomendamos consultar con especialistas en SEO técnico como nuestro equipo.
¿Ofrecen servicios de consultoría personalizada?
Sí, ofrecemos servicios de consultoría SEO técnica personalizada, auditorías completas y optimización integral. Contáctanos para discutir las necesidades específicas de tu proyecto y cómo podemos ayudarte.