On-page SEO is everything on the page itself that helps a search engine understand what it is about and helps a person decide to click it — and roughly half of the advice in circulation about it is built on limits Google does not publish.
Two examples, both from Google's own documentation.
There is no character limit for a title tag. Google says title links are "truncated in Google Search results as needed, typically to fit the device width." Not 60 characters. Not any number.
There is no character limit for a meta description either, and Google states it only uses yours "when it describes the page better than other parts of the content." The rest of the time it writes its own.
The 60-character and 155-character conventions are useful working proxies for will this probably fit. They are not rules, and treating them as rules produces titles trimmed to hit a number rather than to describe a page.
What on-page SEO actually consists of
Six things, in order of how much they change outcomes.
1. The page genuinely answering its query
Nothing below matters if this fails. A page that partially answers the question it ranks for loses to one that answers it completely, and no amount of markup or formatting closes that gap.
2. A title that describes the page and earns the click
The title link is the largest single influence on whether an impression becomes a click. It is also not entirely yours — see the next section.
3. Structure that makes the answer extractable
Answer-first sections, one idea per heading, self-contained passages. This has always been good writing; it now also determines whether a passage can be lifted into a generated answer, covered in content formats that still get clicks from AI.
4. Internal links that give the page context and standing
The most under-used on-page lever, and the one with the largest gap between effort and effect. In internal linking: how to build topic clusters.
5. Images that load fast and are described accurately
Both an accessibility and a performance question before it is a search one. In image SEO.
6. Metadata that is accurate rather than optimised
Title, description, headings, alt text. Accurate beats stuffed, and Google's guidance says so directly in every one of those documents.
The title tag is not fully under your control
This is the part most on-page advice omits, and it explains the frustration of writing a careful title and seeing something else in results.
Google generates the title link from seven sources, of which your <title> element is one:
- The
<title>element - The main visual title on the page
- Heading elements
og:titlemeta tags- Prominent styled text
- Anchor text on your own page
- Text in links from other sites pointing at the page
Two of those are not on your page at all. Which means the title Google shows can be influenced by how other people link to you.
Why Google rewrites a title
Its documentation names the triggers:
- The title is incomplete or out of date
- It inaccurately describes the page
- It is duplicate boilerplate repeated across pages
- The page has multiple prominent headings and the main one is unclear
- The language or writing system does not match the page
The most common of these by far is boilerplate. A title of "Blog | Company Name" on two hundred posts is Google's cue to write something better, because it has no choice.
The practical fix is not a clever title formula. It is: make the <title> and the visible H1 say the same thing, make that thing specific to the page, and stop repeating your brand name in a position where it displaces the description.
Headings
One H1 per page, matching what the page is about. WordPress themes usually handle this; some render the site name as an H1 on every page, which is worth checking once.
Use H2 and H3 to structure meaning, not to size text. A heading exists to tell a reader — and a retrieval system — what the next section answers. Choosing H3 because it looks better is a styling decision made in the wrong place.
Headings are one of the seven title sources. A page with four competing prominent headings gives Google an ambiguity to resolve, and it may resolve it differently from how you would.
Content structure that works for both readers
Four habits, all of which help people and machines for the same reason.
Answer the question in the first two sentences of each section. Retrieval extracts passages. A section that arrives at its point in paragraph four gets skipped in favour of one that did not make the reader wait.
Make each section self-contained. Assume the reader arrived at that heading and read nothing before it. "As we saw above" breaks the passage the moment it is lifted out.
Use tables for anything comparative. They are scannable, quotable as a unit, and hard to paraphrase without losing the point.
Be specific. Dates, figures, versions, named error codes. Specifics are what make a page worth citing rather than summarising, and what a fact-check can confirm.
URLs
Short, descriptive, stable. /blog/on-page-seo beats /blog/2026/09/03/the-complete-guide-to-on-page-seo-for-beginners.
Do not change a URL that ranks unless you must. If you must, redirect it properly with a 301 and expect a temporary dip.
Words, hyphens, lower case. Underscores, capitals and IDs all work and none of them help.
URLs are a weak signal. They are worth getting right at creation and almost never worth changing afterwards — the cost of a migration exceeds the benefit in nearly every case.
What is not worth your time
Keyword density. Not a thing, and has not been for many years.
Exact-match keyword in the first 100 words. Write the sentence that belongs there.
LSI keywords. Not a technique search engines implement. The underlying instinct — cover the subject properly — is right; the vocabulary around it is invented.
Hitting an exact title or description length. No published limit exists. Aim to be complete and not truncated on mobile, which is a design constraint rather than a character count.
FAQ schema for rich results. Retired on 7 May 2026. Detail in schema markup for blogs.
Word count targets. "This page should be 1,800 words" is a competitor average dressed as a requirement. Length follows from covering the subject.
The order to work in
For a new page: query and intent, then the answer, then structure, then title and description, then internal links, then images.
For an existing page that is slipping, the order is different and starts with diagnosis — the method is in how to refresh old blog posts that are slipping.
The check that saves the most wasted work: open Search Console, find the page, and compare impressions against clicks. High impressions with low clicks is a title and description problem. Low impressions is a relevance or authority problem. They need opposite responses, and doing on-page work on the second one changes nothing.
Frequently asked questions
What is on-page SEO?
Everything on the page itself that helps a search engine understand it and helps a person decide to click. In order of impact: genuinely answering the query, a title that describes the page, extractable structure, internal links, fast and well-described images, and accurate metadata.
How long should a title tag be?
Google publishes no character limit. Its documentation says title links are truncated in results as needed, typically to fit the device width. The common 60-character guidance is a working proxy for what usually fits on mobile, not a rule — write a title that describes the page and check how it renders.
Why does Google change my title tag?
Because the title link is generated from seven sources, of which your title element is one — along with the visible title, headings, og:title, prominent styled text, your own anchor text, and links from other sites. Google rewrites when a title is incomplete, inaccurate, duplicate boilerplate, or when the main heading is unclear.
Does the meta description affect rankings?
No. It affects whether someone clicks. Google states it uses your description when it describes the page better than other parts of the content, and otherwise generates its own snippet from the page — so a well-written description is a click-through investment, not a ranking one.
Does keyword density matter?
No, and it has not for many years. Neither do LSI keywords, which are not a technique search engines implement. The useful instinct underneath both — cover the subject properly and use the language your readers use — remains correct.
How many words should a blog post be?
There is no target. Word count guidance is usually a competitor average presented as a requirement. Length should follow from covering the subject completely; a 900-word page that answers the question beats a 2,400-word page that circles it.
Should I change my URLs to be more SEO-friendly?
Almost never, if the page already ranks. URLs are a weak signal and migrations carry real risk, so get them right at creation — short, descriptive, hyphenated, lower case — and leave working ones alone.
How do I know if my on-page SEO is the problem?
Compare impressions with clicks in Search Console. High impressions and low clicks is a title and description problem, which on-page work fixes. Low impressions is a relevance or authority problem, which it does not.
What to do next
Open Search Console, sort your pages by impressions, and look for anything with high impressions and a low click rate. Those pages are already ranking and failing to earn the click, which makes them the cheapest wins you have.
Then check whether your <title> and your visible H1 say the same thing on those pages. When they disagree, Google picks — and it may not pick yours.
Related guides
- Meta titles and descriptions: write them for clicks — the detail on both
- Internal linking: how to build topic clusters — the most under-used lever
- Image SEO: alt text, file size and what ranks — the performance half
- How to refresh old blog posts that are slipping — fixing pages that already exist
- Technical SEO: what actually breaks rankings — the layer beneath this one
- Content formats that still get clicks from AI — structure for generated answers
Chapters in this Guide
Free: The 60-Minute Email Authentication Fix
A no-fluff checklist to set up SPF, DKIM & DMARC correctly and pass Gmail & Yahoo's sender requirements.
Muhammad Basim has worked in digital marketing since 2013, focused on email deliverability and AI-assisted content production. He is the author of The Email Deliverability Playbook and The Email Copywriting Playbook.
Related Articles
Email Deliverability: Why Authenticated Emails Still Land in Spam
Email deliverability is whether your message reaches the inbox rather than the spam folder, and it is decided by three factors: technical infrastructure, list quality, and sending behaviour. Authentication is one item inside the first factor. Getting it right is necessary, and it is nowhere near sufficient. That gap explains the most common complaint in […]
Linkable Assets: What Actually Earns Links
A link is a citation, and a citation requires that someone writing about your subject needed you to make their point. That is the whole mechanism, and it is the reason most "linkable content" earns nothing. The test, before you build anything: Could a writer covering this topic finish their sentence without referencing you? If […]
Link Outreach Emails That Get Replies
Outreach fails for exactly two reasons, and only one of them gets written about. The first is that the email was not worth replying to. That is the reason every guide addresses, and the advice — personalise, be brief, offer value — is correct and insufficient. The second is that the email never arrived. Nobody […]