Why Keyword Stuffing Hurts Your SEO
Nobody repeats a phrase twenty times in a paragraph any more, so a guide describing that is useless. The forms that survive are subtle enough that businesses commit them without knowing, which is what this covers.
What It Actually Is Now
Keyword stuffing is repeating a phrase because you believe it helps rather than because the sentence needed it. That definition covers the crude old version and the subtle current one equally. It puts the test where it belongs, which is on the intention behind the writing rather than on any count.
What it used to look like. Obvious.
Lists of towns at the foot of a page. Text hidden in the background colour. The same phrase repeated until a paragraph was unreadable. All of it stopped working a very long time ago.
Why the old descriptions persist. They are easy to write.
Explaining that hidden text is bad requires no judgement. Explaining that a service page mentions its own subject three times too often requires reading it.
What it looks like now. Almost normal.
Copy that reads slightly wrong. A sentence using a full service name where anybody would have said it. A heading that repeats what the heading above it said.
Who does it. Not cynics.
Businesses following advice they were given. Or writers handed a phrase list. Almost nobody doing this is trying to manipulate anything.
Why it still matters. It costs twice.
Once with the reader who arrived, which block four covers. Once with the assessment of the page itself, which is block five.
Density Is Not A Real Measure
People ask what a good keyword density is. There is no correct figure. We will not print one even as an example of outdated advice, because a number published in order to be rejected still ends up quoted back at us. The question does not describe anything search engines assess.
What density actually is. A measurement of your page.
How often a phrase appears relative to the words around it. Anybody can calculate it. That does not make it something being evaluated.
Where the idea came from. Early matching.
When engines genuinely counted words, the count meant something. It has not for a very long time. The measure outlived the mechanism entirely.
Why tools still report it. It is calculable.
A number that can be produced from a page will be produced and displayed, regardless of whether it corresponds to anything. That is true of several figures in this field.
What a target does to writing. Ruins it.
A writer aiming at a figure produces sentences shaped by arithmetic. The page gets worse while the number goes up, so the exercise appears to have worked.
What to say to a supplier. Ask what it predicts.
If somebody reports your density, ask what outcome the figure predicts and how they know. There is no good answer and the question ends the conversation.
The related myth. Required related terms.
The same instinct dressed up differently, which our guide to LSI keywords covers.
The Local Version
Local businesses stuff place names. It is the single commonest form we find and it is almost always well intentioned. It reads worse than any other version because human beings simply do not talk that way about where they are.
What it sounds like. Unnatural.
A sentence naming the town, then the next sentence naming it again, then a heading with it in as well. Nobody describes their own area that often in one breath.
Why businesses do it. They were told to.
Somewhere along the line every local business is advised to mention their town. That advice is not wrong. It gets applied with no sense of proportion.
What is legitimate. Saying where you are.
Stating your base, listing the areas you cover, mentioning a place when the content is genuinely about it. All useful and all normal.
Where it tips over. Sentences that did not need it.
When the place name appears in a sentence about something else entirely, purely to be present. That is the line and it is usually obvious on reading.
The larger scale version. Templated town pages.
One page duplicated across a dozen places with the name swapped. Our landing pages guide covers where that line sits.
What actually works instead. Local detail.
One mention plus something only somebody working there would know beats six mentions of the name. The specifics do the job the repetition was attempting.
What It Costs You With Readers
Over-optimised copy reads as untrustworthy. Not wrong, not badly written exactly, simply off in a way people notice without being able to name. For a business asking somebody to let a stranger into their home, that impression is expensive.
What the reader actually perceives. Something automated.
They do not think about keyword density. They think this reads like it was produced rather than written, which lands as a business that might not be entirely real.
Why it matters more for trades. Trust is the product.
Somebody choosing an electrician is deciding whether to trust a stranger with their house and their safety. Copy that sounds mechanical works directly against that.
Where the damage happens. Silently.
Nobody rings to say your page read oddly. They go back to the results and call somebody else. Nothing in your reporting explains it.
What good copy signals instead. A person.
Specific detail, plain description of what happens, limits stated openly. All of it says somebody who does this work wrote this, which is the only credential a web page can offer.
The comparison worth making. Read a competitor.
Read your page and the best competing page one after the other. The difference is usually immediate and it is rarely about information.
The uncomfortable truth. Ranking is not the finish.
A page that ranks and reads badly delivers visitors to a bad impression. That is worse than not ranking, since it costs money to achieve.
What It Costs You With Search Engines
The useful framing is not punishment. It is that repetition fails to demonstrate the thing being assessed. A search engine is trying to work out whether a page genuinely answers a search. Saying the same phrase repeatedly is not evidence that it does.
What is actually being assessed. Whether you answered.
Coverage, specificity and whether somebody who made that search would leave satisfied. Repetition contributes nothing to any of those.
Why repetition can actively hurt. It displaces.
Every sentence spent restating the phrase is a sentence not spent answering something. The page ends up shorter on substance than it looks.
What about penalties. The wrong question.
Formal action is possible in egregious cases and it is not what happens to an ordinary business with slightly overworked copy. Our risky practices material covers that subject properly.
The realistic outcome. Quiet underperformance.
The page simply does not do as well as a better written one on the same subject. No notification, no drop, nothing to diagnose.
Why that is harder to spot. Nothing broke.
A penalty would at least be visible. Underperformance looks like a page that exists and never quite works, which businesses tolerate for years.
What fixing it involves. Rewriting, not removing.
Taking the phrase out is not the fix. Writing the page as somebody explaining the job would is the fix. The phrase count sorts itself out.
The Modern Forms
These are the versions we actually find on client sites. None of them looks like the classic description and all of them come from following advice. Most businesses committing them would be surprised to be told.
Every heading saying the same thing. The commonest.
A page where each subheading contains the service name. Individually reasonable. Read in sequence they announce that somebody was optimising rather than organising.
The services list padded with variations. Close second.
Boiler repair, boiler repairs, emergency boiler repair, boiler repair service. Four entries describing one thing, presented as a list of offerings.
The full name where a pronoun belongs. Subtle.
A paragraph that will not say it or the work. Every reference is the complete service name, which no person speaking aloud would do twice in a row.
Alt text written for machines. Frequently inherited.
Image descriptions stuffed with service and place names rather than describing the picture. That text exists for people who cannot see the image. Using it this way fails them.
The introduction that restates the title. Common in inherited copy.
A first paragraph that says nothing except the page subject in slightly different words. The reader learns nothing and the opening is wasted.
Questions written as headings for the phrase. Newer.
A run of headings phrased as questions nobody asks, each containing the target words. That reads as a page arranged around a list rather than around a subject.
How To Tell If You Are Doing It
Three checks, none of them requiring a tool, all of them faster than any report. If a page passes these it does not have a stuffing problem, whatever a checker might say about it.
Read it aloud. The best single test.
Anywhere you stumble is where the writing was bent. Anywhere you would naturally substitute a pronoun and the page has not, somebody was inserting.
Read the headings in order. On their own.
Skip the body and read only the headings. If they repeat the same words, they were written for a machine rather than to help somebody find their way down the page.
Ask whether a sentence earns its place. One at a time.
Take any sentence containing the target phrase and ask what the reader learns from it. If the answer is nothing, it exists for the phrase.
What not to use. A density checker.
It will give you a number and the number means nothing, so acting on it will make the page worse in a measurable direction.
How to fix what you find. Say it plainly.
Rewrite the sentence as you would say it to a customer standing in front of you. That single instruction resolves nearly every instance.
Where to start. The commercial pages.
Fix the pages that produce enquiries before anything else. Everything sits on the keyword research guide.
Your copy reads
slightly wrong
and nobody
rings to say so.
Over-optimised writing does not read as deceitful. It reads as produced rather than written, which for a business asking somebody to let a stranger into their home is expensive. They go back to the results and call somebody else. Nothing in your reporting explains why.
What we look for in inherited copy:
If somebody reports your keyword density, ask what outcome that figure predicts and how they know.
Every guide.
One subject.
What the words mean, what the numbers in a keyword tool actually are, how to find the searches worth targeting, how to turn a list into a publishing plan and what to do when a page you relied on slips.
