The on-page SEO checklist we run on every page, 18 ways
Not "get more backlinks" or "post more content" — the boring, mechanical, entirely-within-your-control stuff that a page either has or doesn't.
Most SEO advice is about things outside the page — links from other sites, domain age, publishing history. All real, none of it fixable this afternoon. On-page SEO is the other half: the 18 things that live entirely inside the page, checkable in one sitting before you hit publish. None is individually decisive; together they decide whether a search engine can make sense of the page at all. (Unfamiliar terms are a sentence each in the glossary.)
Six groups, eighteen checks
Title and meta description
Every page needs its own title — not a placeholder, not the site name repeated on every page — and it reads best between 30 and 60 characters; much longer and search engines cut it off mid-sentence in results. Pair it with a meta description, the short paragraph that often shows under your title in search results: 70 to 160 characters is the range that neither gets truncated nor looks sparse. Skip both and you're leaving search engines (and impatient readers) to guess what the page is about from the URL alone.
One heading structure, not two
A page should have exactly one main heading (H1) — not zero, not three competing for the job — and everything under it should step down one level at a time: H1 to H2 to H3, never H2 straight to H4 because a smaller heading "looked right" in the design. Skipped levels don't just look odd to a screen reader; they make it genuinely harder for a search engine (or a human skimming) to tell what's a section and what's a sub-point.
Say what you want to be found for
Pick the one phrase you want a page to be found for — its focus keyword — and use it deliberately: once in the title, once in the main heading, and at least once more, naturally, in the body text. This isn't keyword-stuffing; it's making sure the phrase you're targeting actually appears on the page you're targeting it with, which sounds obvious until you check a few pages and find it doesn't.
Every image earns its alt text
Every image that carries real content — a product photo, a screenshot, a diagram — needs alt text describing what it shows; purely decorative images (a background texture, a divider) are the one exception and can be marked as such instead. The target is 100% coverage on content images, not "most of them" — screen readers and search engines both treat a missing alt the same way: as if the image weren't there at all. More on writing alt text that's actually useful, not just present, in the dedicated guide.
Enough content, linked to the rest of the site
Aim for at least 300 words of real, visible body content — enough to actually answer what someone searching for the topic wants to know, not padding to hit a number. Then link out: at least two links from the body copy to other relevant pages on your own site. A page that mentions related topics but never links to them is asking a visitor (and a crawler) to go find those pages themselves instead of just following a link that's already there.
6The checks nobody remembers until something breaks
The URL slug: short, lowercase, hyphenated — online-store, not Online_Store_Final_v2. A canonical address, so near-duplicate URLs don't compete with each other. Nothing accidentally flagged "don't index" — common enough to have its own guide. A social share image at 1200×630 or larger, so links posted elsewhere show a real preview. And a title and description that aren't word-for-word copies of another page's — search engines read duplicates as one page with two addresses. (Meta keywords you can skip entirely: no major engine has read it in over a decade.)
What this looks like on one real page
No — each carries its own weight, and a handful (a working title, exactly one H1, not being accidentally hidden from search) matter far more than the smaller ones like the meta-keywords field, which carries no weight at all. How the weighting actually works is its own guide if you want the mechanics.
Some don't — a page with no focus keyword set skips both keyword checks rather than failing them, and a page with no images at all passes the alt-text check by default. A well-built checker should treat "doesn't apply" differently from "fails," and not punish a page for a check that was never relevant to it.
No, and treat anyone who says otherwise with suspicion. On-page checks are the things fully inside your control — they remove the self-inflicted reasons a good page underperforms. Ranking well also depends on demand for the topic, competition, and signals from outside the page entirely, none of which a checklist like this one touches.
Any time you edit it meaningfully, and once as a matter of course before you publish anything new. Titles and descriptions especially tend to drift — a page written for one topic gets repurposed for another, and the metadata quietly stops matching what's actually on the page.
Older pages often benefit more, if anything — they're exactly where titles have drifted, links have rotted, and a focus keyword was never set because the page predates the idea of setting one. A quarterly sweep across pages you already have usually turns up more easy wins than agonising over one page that's still being drafted.