How to Optimise Existing Content for SEO
A page already appearing somewhere has proved two things: the search exists, plus your site can compete for it. That makes improving it a considerably shorter route than starting again. It is also the work most businesses skip in favour of writing something new.
Improving Beats Adding
An existing page that appears anywhere has already settled the two questions a new page has to answer from nothing: whether anybody searches for this, plus whether your site is considered a plausible source.
What a new page has to establish. Everything.
That the subject is searched, that your site can be found for it, that the page is worth showing. All from a standing start, over months rather than weeks.
What an existing page has already established. Most of it.
It is known, it is being considered and something has decided it is relevant enough to show. The remaining distance is smaller than it looks.
Why businesses skip this. New feels productive.
Publishing something new is visible and satisfying. Improving something already published is invisible, so it feels like standing still even when it returns more.
Which pages to improve. The ones close to earning.
Pages appearing but not high enough to bring anybody. That is the shortest available distance in on-page work. The audit page covers identifying them.
What improvement is not. A rewrite.
Replacing a page wholesale discards whatever it had established and starts the clock again. The point is to keep what works and close what is missing, which is a smaller and more careful piece of work than it sounds.
The boundary worth stating. Posts specifically.
Updating old blog posts as an ongoing discipline is covered in our blogging material. This page covers improving any page against a search it already targets.
Start By Comparing Against What Ranks
The useful question is what the current results cover that your page does not. That is judgement rather than a tool output. It is the single most productive half hour available on an existing page.
What to do. Read the competition properly.
Open what currently appears for your search and read it as a customer would. Not scanning for phrases. Establishing what it covers and where yours is thinner.
What you are looking for. Gaps.
Questions they answer that you ignore, distinctions they make that you skip, information they provide that you assumed nobody needed. Those are your list.
What you are not looking for. Their structure.
Copying somebody's arrangement produces a similar page rather than a better one. The point is to find what is missing from yours, not to reproduce theirs.
How long it takes. Less than writing anything.
Reading four competing pages properly is an hour at most. It produces a specific list of what to add. Compared with writing a page from nothing, that is a small investment against a much better informed piece of work.
Why no tool does this. It requires reading.
Software can compare terms and produce a list of words you lack. It cannot tell you that everybody explains a step you omitted, which is the finding that matters.
The second thing it tells you. Whether to continue.
Sometimes the comparison shows the current results are genuinely better and you cannot realistically match them. Learning that in half an hour is worth the half hour.
Content Depth
Depth means covering what the reader needs. It does not mean reaching a length, because length is not a ranking factor. Padding a page to look substantial makes it worse rather than deeper.
What depth actually is. Completeness.
Somebody finishing your page should not need another one to act. That is the test. It is unrelated to how many words getting there took.
Why word count persists as advice. A correlation misread.
Pages covering subjects thoroughly tend to be longer, so somebody measured length and concluded length was the cause. The relationship runs the other way.
What padding does. Dilutes.
Adding preamble, repetition and tangents to reach a target buries the answer further down. That costs you the reader who arrived with a question, which is the whole audience.
What thoroughness actually costs. Editing down.
A page covering a subject completely and briefly is harder to write than a long one, because it requires deciding what to leave out. Length is frequently the residue of not having made those decisions.
Where the opposite failure sits. Genuinely thin pages.
A page that does not answer its question is a real problem. The fix is answering it rather than lengthening it. Those look similar and are entirely different.
How to tell them apart. One question.
Ask whether a reader could act after finishing. A short page answering completely is deep. A long page that never quite answers is thin, however substantial it looks. Length is what disguises the second case.
Why we state no target. There is not one.
No figure exists that would be correct for every subject. Publishing one invites writing towards it. You will not find a length target anywhere on this site.
LSI Keywords
Latent semantic indexing is a real information retrieval technique from decades ago. LSI keywords, as the industry uses the term, is not something search engines work with. Google has said so directly on more than one occasion.
What the real technique is. Genuine and old.
An approach originating in a research paper from 1988, designed to analyse relationships between terms across a small fixed collection of documents. Real mathematics, built for a library rather than the web.
What the industry term claims. A list to include.
That search engines look for particular related terms, then that including them from a supplied list demonstrates coverage. That is a claim about mechanism rather than about writing well.
What Google said. Twice, publicly.
On 30 July 2019, Google's John Mueller stated there is no such thing as LSI keywords
and that anybody saying otherwise is mistaken. In January 2023 he repeated it, noting that anybody recommending them remains wrong.
What researchers found. The same conclusion.
The late Bill Slawski, who worked extensively through Google's patents, made the point that what the industry calls LSI keywords neither use the technique nor are keywords. Two independent routes to one answer.
What the tools are selling. A name.
Generators offering LSI keyword lists are named after a technology not in use. The lists themselves are related terms, which is a real thing described by a false label.
Why the term survives. It sounds technical.
An acronym referencing real mathematics is more persuasive than write about the subject properly. That is the whole of its appeal. It is why the term has outlived two direct corrections.
Related Keywords
Covering a subject properly naturally produces related language. That is the real observation the myth was built on. It is worth keeping once the mechanism claim is discarded.
What is genuinely true. Vocabulary follows subject.
Anybody explaining something uses the words belonging to it, because those are the words for the things involved. A page covering a subject contains that language without anybody inserting it.
Why that got mistaken for a technique. The correlation is real.
Good pages do contain related terms. Somebody observed that and concluded the terms produce the quality, which reverses cause and effect and sells better.
How search engines actually handle this. Understanding, not matching.
Modern systems assess whether a page covers a subject, using context and relationships between concepts. That is genuinely semantic and it is not a list of terms to include.
What to do with a related terms list. Read it for gaps.
A list of related terms is useful as a prompt: if a term you have never mentioned belongs to the subject, you may have missed something. That is research rather than insertion.
Why the distinction matters commercially. One is buyable.
A list of terms can be purchased and delivered. Covering a subject properly cannot, which is why the first gets sold as a service and the second gets described as obvious.
Where this argument continues. The semantic material.
Our advanced guides cover how subject coverage is assessed rather than term matching, which is the fuller version of this block.
Content Freshness
Recency matters for subjects where recency matters, though not for subjects where it does not. Changing a date without changing the content achieves nothing. It is a small deception that is easily detected.
Where recency genuinely matters. Moving subjects.
Anything where the answer changes: rules, prices, product versions, guidance that has been updated. A stale page on those is actively misleading rather than merely old.
Where it does not. Settled subjects.
How something works, what a term means, principles that have not changed. A page from four years ago explaining a stable concept is not worse for being four years old.
What updating a date does. Nothing.
Changing a displayed date while leaving the content untouched does not make the page fresher. It makes the page misrepresent itself, which is a different thing entirely.
Why it is detectable. The content betrays it.
A page dated recently that references superseded rules or discontinued products is obviously not current. A reader notices. So does anything assessing the page.
What genuine updating involves. Actually revising.
Reading the page against current reality and changing what is no longer true. That earns a new date. Our blogging material covers doing it as a routine.
What Actually Moves A Page
Four things, ordered by how much difference each tends to make. The order matters more than the list, because most effort goes to the wrong end of it.
Matching intent properly. The largest.
Serving the want the search actually carries. A page aimed at the wrong intent cannot be improved into working, which search intent explains.
Covering what is missing. The second largest.
Adding what the comparison in block two revealed. Genuine gaps rather than more words, which is the difference between depth and padding.
Clearer structure. Third.
Headings that describe the argument and an order somebody can follow. Frequently the cheapest of the four and rarely the reason a page is failing.
Internal links pointing at it. Fourth and underrated.
Links from related pages on your own site. Genuinely effective, entirely within your control and almost always neglected in favour of editing the page itself.
What is missing from the list. Anything technical.
None of the four is a setting, a tag or a count. On a page that is already appearing, the remaining distance is almost always about the reader rather than about the markup, which is worth knowing before commissioning anything.
Why the order is the point. Effort goes to the wrong end.
Most improvement work is structural tidying and word changes, which sit third. The first two require thinking about the reader, which is harder and returns more.
Know When To Stop
A page can be improved indefinitely with diminishing return. Once the obvious gaps are closed, the effort is worth more on the next page. Recognising that is a skill rather than laziness.
What endless improvement looks like. Refinement without a hypothesis.
Rewording, reordering and adding detail because the page could always be better. Each change is defensible and none is answering a question about why it is not performing.
The stopping test. Two questions.
Have the gaps you identified been closed, plus can you name what the page still lacks. If the gaps are closed and you cannot name a lack, you have finished.
Why stopping returns more. Portfolio arithmetic.
Ten pages improved from poor to good beats one improved from good to excellent, because the first change crosses a threshold and the second refines past it.
Why it feels wrong to stop. The page is still imperfect.
Every page can be bettered, so stopping means leaving something you can see. That discomfort is what keeps people refining. Resisting it is the difference between improving a site and improving a page.
When to come back. On evidence.
Return to a page when something changes: the results shift, the subject moves or the page stops performing. Not on a schedule and not because it could be better.
What this connects to. Prioritising properly.
The audit page covers ordering work across a whole site, which is where stopping decisions actually get made. The full series is on the on page SEO guide.
Improve the page
you already
have.
A page already appearing anywhere has proved the search exists and your site can compete for it. A new page has to establish both from nothing, over months rather than weeks. It is the shortest distance available in on-page work.
In order of effect:
Ten pages improved from poor to good beats one improved from good to excellent.
Every guide.
One practice.
What a page says about itself, headings, keywords, search intent, images, page types, user behaviour, credibility and how to review an existing site.
