On page SEO: the checklist we use on every build
On page SEO is not a plugin and not a dark art. It is a set of ordinary decisions about structure, content and markup, made during the build rather than after it.
Contents
On page search work has a poor reputation because it is usually sold as a separate service, applied after a site is built, by someone who was not in the room when the decisions were made. Done that way it consists of filling in fields and hoping.
Done properly it is part of the build, it takes very little extra time, and most of it is indistinguishable from simply making the site clear. Here is the list we work through.
Structure, before anything else
Search engines and readers both need to understand what a site covers and how its parts relate. That is an information architecture problem, and it is settled at the start.
- One page per thing you want to be found for. A single services page listing eight services will not compete with eight pages that each address one.
- A shallow hierarchy. Anything important should be reachable in two or three steps from the home page.
- Stable, readable URLs that describe the page. No dates in a URL for content that is not news. No identifiers where a word would do.
- No two pages competing for the same intent. If they do, merge them. Two mediocre pages beat neither one good page.
Titles and descriptions, written not generated
The title element is the single most important piece of on page markup. It is what appears in results, in browser tabs and when a link is shared.
Write one per page, by hand. Put the specific thing first and the business name last. Keep it short enough not to be truncated. Make it read like a sentence a person would say rather than a list of keywords separated by pipes.
The meta description does not affect ranking directly, but it affects whether anyone clicks, which affects everything downstream. Treat it as advertising copy with a strict character budget: one sentence that tells the reader what they will get from the page.
Headings that describe the page, not the design
One h1 per page, saying what the page is. Then h2 for each major section and h3 beneath those, nested properly, with no levels skipped for visual reasons. If a heading looks wrong at the right level, that is a styling problem, and styling is what CSS is for.
This is also an accessibility requirement, and the two goals point the same way often enough that we treat them as one job.
Content, which is the actual work
The parts of on page SEO that are ticks on a list take an hour. The part that decides whether it works is whether the page is the best answer to the question somebody asked.
- Answer the question in the first paragraph. Do not build up to it.
- Use the words your customers use, not the words your industry uses internally.
- Cover the follow up questions on the same page. Someone who has to search again did not get what they came for.
- Be specific. Specifics are what distinguish your page from the generic version of it that already exists a thousand times.
- Say who you are and where you are, on pages where that is relevant. Local intent is extremely common and frequently unserved.
Internal linking
Links between your own pages tell search engines which pages matter and give readers a route onward. Link from the general to the specific and back again. Use descriptive link text that makes sense read aloud on its own, which rules out “click here” and “read more” as the entire link.
Every page should be reachable from somewhere other than the sitemap. An orphaned page is a page you have decided not to promote.
Images
Alternative text that describes the image for someone who cannot see it. Not a keyword list, a description. Decorative images get an empty alt attribute so screen readers skip them. Filenames that describe the contents. Correct dimensions and modern formats, which is a performance concern and a ranking one.
Structured data
Mark up what the page genuinely is: an organisation, a local business with real opening hours, an article with a real author and date, a product with a real price and availability. Structured data helps machines understand the page, and increasingly it is how assistants and answer engines resolve which entity a page belongs to.
Two rules. Only mark up things that are actually on the page. Never mark up a rating or a review you do not have, which is both a policy violation and a lie.
The technical floor
None of the above works if the page cannot be crawled or does not load. The minimum:
- A canonical URL on every page, so duplicate routes to the same content do not compete.
- One version of the site: one protocol, one host, everything else redirected permanently.
- A sitemap that contains what you want indexed and nothing else, and a robots file that does not accidentally exclude something important.
- Redirects from every old URL when a site is replaced. This is where the most damage is done and it is entirely avoidable.
- Pages that load quickly on a phone, and do not move about while they load.
- Working, distinct pages rather than content assembled by script after the page arrives, unless you are certain that is being rendered correctly.
What we do not do
We do not buy links, spin content, stuff keywords into footers, or promise a position for a term. Nobody can promise a position, and anyone who does is either guessing or planning to pick a term nobody searches for so the promise is technically kept.
What we do is build the site so the fundamentals are right from the first day, because retrofitting structure onto a finished site costs several times what doing it correctly costs during the build, and some of it cannot be retrofitted at all.