On-Page SEO:
The Fundamentals

On-page SEO is the work you do on a page to help it rank. It is concerned with the content itself, plus the HTML, structure, and internal links around it. It is also the only part of SEO you can act on today without anyone’s permission. You don’t need to do outreach, no need for a developer sprint, or waiting for a domain to earn trust.

This guide covers the entire surface of on-page SEO, ordered by the impact they can make. Some changes can move rankings on its own after being implemented, some compounds across a site, and the rest is hygiene you finish once and stop revisiting. We’re here to explain which is which, so you can decide where to put your hours in.

For the wider context, our step-by-step SEO guide covers how on-page fits alongside the rest of the work.

What On-Page SEO Covers (and Where It Stops)

If you can change it by editing the page, it’s on-page SEO. That gives you a fairly short inventory:

• The title tag, which tells search engines and searchers what the page is about.

• The meta description, which shapes how the result reads in the SERP.

• Headings, H1 through H6, which organize the page and signal its structure.

• The URL slug, which describes the page and reflects where it sits on your site.

• The body content, meaning the depth, accuracy, and originality of what you wrote.

• Intent match, meaning whether the page answers the question the searcher asked.

• Internal links, both the ones pointing out of the page and the ones pointing at it.

• Images, including filenames, compression, and alt text.

• Structured data, the schema markup that makes a result eligible for rich features.

That is the whole surface, and the rest of this guide is about which of them deserve your attention.

Why it comes first

On-page work earns priority for three key reasons.

It is the only SEO you control outright. Backlinks depend on other people, technical fixes depend on developers, and rankings depend on Google. The page itself depends entirely on you, which means it is the one lever available on a Tuesday afternoon with no budget approval.

It is testable. Change one page, watch what happens to it in Search Console, and then learn something from it. Most SEO work resists that kind of clean attribution, and on-page work doesn’t.

And it gates everything else. Links pointing at a page that answers the wrong question don’t rescue it. Technical work that makes a thin page crawlable produces a crawlable thin page. On-page quality is the thing the rest of your SEO investment multiplies, which is an unforgiving relationship when the multiplier is near zero.

Where it stops

The boundary matters because two neighboring disciplines get folded into on-page constantly.

Off-page is everything that happens elsewhere. Backlinks, brand mentions, reviews, coverage. You influence it, you do not control it.

Technical SEO is site-wide infrastructure. Crawling, indexing, rendering, site speed, sitemaps, canonicals. It decides whether your page can be read at all, which makes it a prerequisite rather than a subset. Our technical SEO guide handles that layer.

Page speed shows up in both lists, but so does structured data. The useful distinction lies in the scope. Technical work fixes a template or a server setting and affects thousands of URLs, while on-page work fixes one page. When a problem repeats across your whole site, treat it as technical. When it lives on one URL, treat it as on-page.

The Three Tiers of On-Page Work

Every on-page tactic falls into one of three tiers.

Tier 1 – The work that moves rankings

Three things carry most of the outcome.

• Intent match. Whether the page answers the query the searcher asked. Get this wrong and nothing below it matters.
• Content depth and originality. Whether the page says something the ten pages above it do not.
• The title tag. The strongest single HTML element you control, and the one most likely to change a position when you rewrite it.

If you only ever do three things to a page, do these. They are also the three that take real thought, which is why they’re often buried among twenty items that take five minutes each.

Tier 2 – The work that compounds

These do not move a page on their own. Applied consistently across a site, they compound into something that does.

Internal linking. Heading structure. URL slugs. Content freshness. Semantic coverage of the entities and subtopics a query implies.

The pattern is that each one is worth almost nothing on a single page and quite a lot across three hundred. Judge them at site level, not page level.

Tier 3 – Hygiene, and what you can skip

Do these once, do them correctly, and then stop revisiting them. Alt text. Image compression. Outbound links. Schema markup on the templates that support rich results. HTTPS. Mobile rendering.

And several widely recommended tactics return close to nothing for ranking purposes. Knowing which ones frees up hours:

• Alt text is an accessibility feature first. Write it because screen reader users need it and because it helps image search. Do not expect it to move a page in text results.

• Meta descriptions do not affect rankings. They affect click-through rate, which is a different and still worthwhile goal.

• Keyword density is not a metric. There is no ratio to hit. Cover the topic properly and the words appear on their own.

• Meta keywords have been dead for over a decade. If your CMS still has the field, ignore it.

The point is not that hygiene work is worthless. It is that hygiene work is finite. Finish it, then spend the recovered time on Tier 1.

Start With Intent, Not the Checklist

The most expensive on-page mistake is optimizing a page that was never going to rank, because it answers a different question than the one being asked.

Search intent falls into four rough categories. Informational queries want an explanation. Commercial queries want a comparison before a purchase. Transactional queries want to buy or sign up. Navigational queries want a specific destination. Each maps to a page type, and a mismatch between them is not something a better title tag can rescue.

The diagnostic takes two minutes. Search your target query, then look at what already ranks. If the first page is all listicles and yours is a product page, you have an intent problem rather than an optimization problem. If the first page is all tools and yours is an article, same answer. Google has already told you what it believes the query wants.

Get this right at the keyword research stage and most on-page work becomes refinement instead of rescue. Get it wrong and you can execute every remaining item in this guide perfectly and stay on page three.

One more intent trap worth naming. A single page can only serve one intent well. When you try to make an article rank for both “what is X” and “best X software,” you usually get a page that half-satisfies both and wins neither.

Title Tags and Meta Descriptions

These two get grouped together constantly and behave completely differently.

Writing a title that earns the click

The title tag is the strongest on-page signal you control and the element most likely to change a ranking when you rewrite it. It tells Google what the page is about and tells a human whether to click.

Put the target term near the front, describe what the page delivers, and stop. Clever titles that hide the subject cost you both the ranking and the click. A title reading “Trail Running Shoes: 12 Picks Tested Over 300 Miles” beats “Our Favorite Shoes for the Trail” because it names the topic, the format, and the evidence.

Google does not always keep what you write. An Ahrefs study found the search engine rewrites title tags roughly 61.6% of the time, usually when the title is too long, keyword-stuffed, or disconnected from the page content. That is not a reason to skip the work. Google’s own documentation on title links explains that a clear, accurate, page-specific title is what keeps it from being replaced.

Length is pixels, not characters

Everyone repeats “keep it under 60 characters.” That rule is a rough proxy for the real constraint, which is pixel width. Titles truncate at roughly 600 pixels, and a capital W eats far more of that budget than a lowercase i.

The practical consequence is that a title full of wide characters can truncate at 52 characters while a narrow one survives past 65. Preview your titles rather than counting them, and front-load the words that must survive a cut.

Meta descriptions are a click lever

Meta descriptions are not a ranking factor and have not been for years. What they do is influence whether someone clicks your result over the one above it, and Mangools reports pages with meta descriptions earning around 5.8% more clicks than pages without.

Google uses what you wrote roughly 37.22% of the time by Ahrefs’ measurement, generating its own snippet from page content the rest of the time. Google’s snippet documentation confirms it picks whatever it judges most relevant to the query.

So write them properly for the pages where clicks matter most, keep them specific, and do not spend an afternoon on the description of a page that gets forty impressions a month.

Content That Answers the Query

Content quality is where guides get vague, so here are three things concrete enough to act on.

Information gain

Ranking pages do not simply cover the topic. They add something the existing results do not have. That addition is what earns the position, and it can be modest. Original data from your own accounts, a process you run, a worked example, a photograph of the thing, a correction to a claim everyone else repeats.

Assembling a summary of the top ten results produces a page that is already redundant on publication. Google’s helpful content guidance describes the same test in different words, asking whether the page adds real value compared with other results.

Showing experience on the page

E-E-A-T stands for experience, expertise, authoritativeness, and trustworthiness. It is not a score you can check, and no plugin outputs it. What you can do is put the evidence on the page where a reader and a model can both see it.

That means a named author with a real bio and credentials that fit the topic. Dates that reflect genuine updates. Sources cited and linked. First-hand detail that only someone who did the work would include. Contact information and clear ownership on the site.

For anything touching health, finance, or safety, treat this as the highest priority on the page rather than a finishing touch.

Where to put your target terms

Keyword placement is simpler than the old rules suggested and still worth being deliberate about.

Use the term in the title tag, the H1, somewhere in the first hundred words, and in at least one subheading where it fits the sentence. That is enough for a search engine to understand the page. Everything past that is diminishing.

What matters more is the vocabulary around it. Search engines interpret topics through related terms and entities rather than through repetition, so a page about on-page SEO that never mentions title tags, canonical tags, internal links, or search intent reads as thin no matter how often the exact phrase appears. Cover the concepts a knowledgeable reader would expect and the semantic coverage takes care of itself.

Two failure modes to avoid. Forcing an awkward phrase into a heading because it matches a keyword tool exactly, and writing four pages that all target the same term. The second one is cannibalization, and it costs more than a missing keyword ever will.

Freshness that means something

Content decays. Rankings slide on pages nobody has touched in two years, without any penalty involved, simply because the results around them got better.

Review high-value pages every six to twelve months. Update the parts that changed, remove what is obsolete, add what has emerged since. Changing a date without changing the content is a tactic that models and readers both see through. A structured content audit turns this from a vague intention into a scheduled task.

Structure: Headings, URLs, and Scannability

Structure is how both a reader and a machine work out what a page contains without reading every word.

Header hierarchy

One H1 per page, describing the page. H2s for the main sections, H3s for subsections beneath them, and no skipped levels. The hierarchy should make sense read on its own. If someone read only your headings, they should understand the argument.

Write headings as descriptions rather than labels. “Why filters multiply your URL count” tells a reader something. “Filters” does not.

URL slugs

Short, readable, containing the target term, mirroring your site structure. A path like /seo/on-page-seo works. Something like /blog/post?id=48271 does not.

Set them once at publication. Changing a slug on an established page means a redirect and a temporary ranking wobble, so the gain has to justify that. Leave old URLs alone unless they are genuinely broken.

Formatting for extraction

This is where structure starts paying twice. A page organized into clear question-and-answer blocks gets pulled into featured snippets, People Also Ask boxes, and AI answers.

The format follows the snippet type. Paragraph snippets want a direct forty to sixty word answer immediately under a question heading. List snippets want an ordered or bulleted list with a clear lead-in. Table snippets want an actual HTML table.

The technique is the same in each case. Ask the question as a heading, answer it in the first sentence beneath, then expand. Burying the answer in the fourth paragraph gives an extraction system nothing to lift.

Internal Linking

Internal linking is the most under-used lever on this list, and one you control completely.

Links inside your site do three jobs at once. They pass authority from pages that have it to pages that need it. They tell search engines which pages relate to which topics. And they help a reader find the next thing, which is the reason worth caring about most.

Two categories behave differently. Structural links come from navigation, breadcrumbs, and category listings, and they establish hierarchy. Contextual links sit inside body copy, and they carry more weight because the surrounding text tells a search engine what the destination is about.

A few rules that hold up:

• Link from strong pages to the ones that need help. Your best-performing posts are an asset you can spend.

• Write anchor text that describes the destination. “Click here” wastes the signal.

• Aim for a handful of contextual links per page, not thirty. Two to five is a reasonable working range for most articles.

• Fix orphans. A page with no internal links pointing at it is a page you have asked search engines to ignore.

Organizing this deliberately is what turns a pile of posts into a topic cluster. Our guide to content pillars covers the structure that makes internal linking systematic instead of ad hoc.

Images, Schema, and Outbound Links

The hygiene tier, handled in one pass.

Images. Compress before upload and serve modern formats, since WebP usually cuts file size by a quarter to a third against JPEG. Use descriptive filenames. Write alt text that describes the image for someone who cannot see it, which is its actual purpose. Add width and height attributes so the layout does not shift while the page loads.

Schema markup. Structured data does not lift rankings on its own. It makes your result eligible for rich results, which changes how the listing looks and how often it gets clicked. Article, FAQPage, HowTo, Product, and Review cover most cases, and Google’s structured data documentation lists what each one supports. Implement it at the template level so every page of a type inherits it, and put it in the server-rendered HTML rather than injecting it with JavaScript.

Outbound links. Linking to credible sources costs you nothing and supports the claims you make. Two or three per thousand words is a sane working range. Link to original sources rather than to somebody else’s summary of them, and check them periodically, since dead outbound links accumulate quietly.

Page experience. Core Web Vitals sit closer to technical SEO than on-page, and the targets are worth knowing anyway. Largest Contentful Paint at 2.5 seconds or less, Interaction to Next Paint under 200 milliseconds, Cumulative Layout Shift below 0.1. Google’s Web Vitals documentation covers the measurement. Fix these at the template level and every page benefits at once.

On-Page SEO for AI Answers

A growing share of your audience is now a model reading your page on someone’s behalf, and that changes which on-page work pays.

The reassuring part is that almost nothing here becomes obsolete. Clear headings, direct answers, accurate structured data, and content that says something original are the same instructions that always applied. Machines reward them more mechanically than humans do.

Three adjustments are worth making.

Answer before you elaborate. Retrieval systems extract passages, not pages. A page that states its answer plainly in the second sentence gets quoted. A page that warms up for four paragraphs gets skipped in favor of one that does not.

Write self-contained sections. Each H2 block should make sense lifted out of the page, because that is how it will be read. Pronouns referring back three sections lose their meaning once a passage is extracted.

Put your content in the HTML. The crawlers behind ChatGPT, Claude, and Perplexity do not execute JavaScript. Content that only appears after a script runs is invisible to them regardless of how well it is written. This is where on-page and technical overlap, and where a well-optimized page can still be unreadable.

Worth keeping in perspective. On-page work makes you quotable. It does not make you credible on its own. Our piece on whether backlinks matter for AI covers the other half of that equation.

On-Page SEO by Page Type

On-page advice tends to assume every URL is a blog post. Your site is not made of blog posts, and the emphasis shifts by template.

The failure mode is applying blog-post tactics everywhere. A category page stuffed with 1,500 words of copy below the grid solves nothing, and a service page written like an explainer buries the reason it exists. If location pages are your focus, our guide to local SEO covers that template properly.

When On-Page Isn’t the Problem

Sometimes the honest answer is that the page is fine and more optimization will not help. Four situations where you should stop.

The intent is wrong. Covered above, and worth repeating because it is the most common. If your page type does not match what ranks, rewriting the title is rearranging furniture. Rebuild the page around the right intent or target a different query.

The gap is authority. A page sitting at position fourteen with clean on-page work, matched intent, and genuine depth is usually telling you the domain is not trusted enough for that query yet. That is an off-page problem, and the fix is to build quality backlinks rather than to add another heading.

Two of your pages are competing. When several URLs target the same query, they split signals and neither ranks. Optimizing one harder makes the split worse. Consolidate them, or clearly separate what each one targets.

The page cannot be read. If content depends on JavaScript that crawlers do not execute, or the URL is blocked, or a canonical points somewhere else, the on-page work is invisible. Google’s guidance on consolidating duplicate URLs covers the canonical case. Confirm the page is readable before you optimize it.

There is also a point where a page is simply finished. When intent matches, the content is genuinely useful, the title is clear, and the structure is clean, further tweaking returns very little. Move to the next page.

How to Tell If It Worked

Because you are changing one page at a time, the before-and-after is unusually clean. Use it.

Search Console is where the evidence lives. Before you touch a page, note its average position, impressions, clicks, and click-through rate for its main queries over the previous 28 days. Make the change, then compare the same window four to six weeks later once the page has been recrawled and re-evaluated.

Read the numbers by what they tell you:

• Position moved, impressions followed. The content or title change worked on relevance.

• Position held, click-through rate rose. Your title and description are doing their job even though ranking did not shift.

• Impressions rose, clicks did not. You are surfacing for more queries without matching their intent. Look at which queries.

• Nothing moved at all. Four to six weeks in, that is a signal the problem sits outside the page. Go back to the section above.

Two habits make this reliable. Change one thing at a time on pages you care about, so you can attribute the result. And keep a simple log of what you changed and when, because six months later nobody remembers which title rewrite preceded which jump.

Rankings are also a lagging indicator. Impressions and query counts move first, which is why our guide to search visibility is worth reading alongside a position tracker.

Conclusion

On-page SEO has a reputation as the fiddly part of the job, and that reputation comes from treating twenty-five tasks as though they were equal. A page that answers the right question with something worth reading, under a title that describes it plainly, is doing most of the available work. The rest is maintenance.

That ordering is unpopular because the bottom of the list is the comfortable part. Compressing images and writing alt text feels productive and finishes cleanly. Deciding that a page targets the wrong intent means admitting the last six months of work on it were misdirected. The first costs an afternoon and returns almost nothing. The second is uncomfortable and returns the position.

So spend the time accordingly. Fix intent, write something the results above you do not have, and title it honestly. Then finish the hygiene, log what you changed, and move to the next page. The pages that hold their rankings for years are rarely the most optimized ones. They are the ones that answered a real question better than the alternatives, and never needed rescuing afterward.

Frequently Asked Questions (FAQs)