On-page SEO, element by element

What is on-page SEO, and what is not?

On-page SEO means the things you decide inside the editor of that one page. Server speed, URL structure, sitemaps and redirects are a different job, and its name is technical SEO. The links other sites give you sit outside your hands as well.

This separation is not a matter of vocabulary. Most on-page checklists mix a few technical items in, and the result is a content writer blaming themselves for work they never had access to.

The list of what is genuinely yours is short: the title tag, the meta description, the main heading and the subheadings, the text itself, the images with their file names and alt text, the internal links and their link text, and the URL. Eight things. If you want the same subject in checklist form to tick off at publish time, the on-page SEO checklist was written for that; this lesson explains why each item is on it.

The title tag: the thing people see in the result

The title tag is not visible on the page. It lives in the page code, and its job is to be displayed in the Google result, in the browser tab and in social sharing. The main heading, the one a reader sees above the text, is a different thing and does not have to match it.

Google itself says it does not always use the title tag and may generate a different title based on what is in the page. So the title is a strong proposal, not an instruction. The best you can do is make the proposal accurate enough that replacing it makes no sense.

Three practical rules. Put the main phrase near the front, because the end of a title gets cut off on mobile. Write the title for a click and not for a crawler: the number in it, the year in it, or the way it differs from the other results is what wins the click. And write a unique title for every page; two identical titles mean you do not know what separates those two pages either.

Revisit one habit as well: the site name at the end of every title. If your brand is known, those few characters build trust and earn their space. If it is not, they only took the room where your most useful words would have gone, and on mobile they are the first thing cut.

The figure everyone quotes as the correct length, around sixty characters, is a display convention and not Google policy. Google has published no character limit; what actually exists is the pixel width that fits in the result.

What does the meta description do, and not do?

The meta description does not affect ranking. We have known this for years and someone still asks every week why filling it in changed nothing. What it does is this: it is the text under the title in the Google result, the two lines that decide whether the user clicks you or the next result.

Google frequently replaces that text with a piece of the page itself, especially when the user's query matches something in the body better. Do not read that as a failure; it means your page had the answer.

Leaving it empty is a choice too, not an automatic mistake. If it is empty Google takes a piece of the page itself, and on informational pages that is often not a bad pick. On a page that sells, do not take that risk: there, the two lines under the title are what the user compares between you and a competitor.

So how do you write it. One sentence that states the thing the user wants, not an introduction to yourself. A phrase like "a complete and comprehensive guide about" gives nobody anything. Say instead what specifically is inside the page: which number, which comparison, which answer. And if it is a sales page, write the thing that is there and is not anywhere else.

Arrange the headings for the reader

A page has one main heading and it names the subject of the page. The subheadings under it each open one section, and their order is the order in which a reader asks their questions.

The common mistake is choosing headings for the search engine: an abstract noun phrase like "the importance of on-page SEO" that nobody ever types. A good heading is the question already in the reader's head: "how many H1s should I have?" or "how do I write alt text?".

Do not skip levels. If you jump from a second level heading straight to a fourth because the font is smaller, the structure of the page is lost; font size is a styling job, not a heading job. And several headings in a row with the same grammatical shape, all of them starting with "how to" for instance, make the page read as machine made. Vary the shape.

Who are you writing image alt text for?

Alt text is written for someone who does not see the image: a user reading with a screen reader, or a browser where the image did not load. That Google reads it too is a side effect, not the purpose.

So write it properly: a description of what is actually in the image, in a short phrase. If the image is a chart, say what the chart shows rather than writing the word "chart". And if the image is purely decorative, leave the alt text empty; empty is a correct choice, not a forgotten field.

Two other things get done at the same moment and matter more than the alt text. First the file name: a descriptive name instead of a string of digits. Second the image width and height: if they are not in the page code, the browser does not know how much room to keep and the text jumps when the image arrives. That jump is one of the three main page experience metrics, and unlike much else it is entirely in your hands.

Internal links: the cheapest work most people skip

An internal link is a link from one page of your own site to another page of the same site. It costs nothing, needs nobody's permission, and in practice gets done less than any other task.

It does two things. First, it tells Google which of your pages matter and how they relate. Second, and more importantly, it moves a reader from the page where they learned something to the page that does the job. That is the commercial path of an educational site, with no hard sell anywhere in it.

Choose the link text carefully. "Click here" tells nobody what the destination is, neither the user nor Google. The link text should describe the target page, and for one given page it should vary a little each time. A hundred links with exactly identical text is a pattern you built yourself.

One working habit makes a real difference: the day you publish a new page, link to it from three older related pages that same day. A page nothing else points at does not exist as far as the structure of the site goes, even if it is in the sitemap. This takes five minutes and almost nobody does it.

One technical detail that looks simple and gets broken constantly: a link has to be an a tag with a real address in href. A button that moves the user with JavaScript works for the user and is not a link for a crawler. Google's own documentation says precisely this.

The URL: get it right once, then leave it alone

A good URL is short, separated with hyphens, and readable enough to tell you what the page is about. Do not put a date in it unless the page really is news, because that page will look old next year without its content being old.

The main point of this section is a warning rather than a guideline. Do not change the URL of a page that ranks. All of that page's history is attached to that address, and changing it means starting from zero unless a permanent redirect from the old address to the new one goes in at exactly the same moment.

And if you do have to change something, change one thing at a time. Do not move the title, the URL and the text on the same day; if the ranking shifts afterwards, you will never know which one caused it.