Writing SEO copy without sounding like SEO

Why keyword density is no longer a criterion

The idea that a term has to reach a certain frequency comes from a time when search engines counted words. That is long over.

Today's systems capture meaning. They recognize that having a website built, web design agency and getting a homepage made refer to the same field, without all three appearing in the text. They also recognize whether a text answers a question or writes around it.

What counts instead:

Completeness. Does the text answer the question and the obvious follow-up questions?

Recognizable expertise. Does it contain details only someone who has actually done the thing knows: numbers, processes, exceptions?

Structure. Can you skim it and still understand it?

The practical consequence: if a term appears unnaturally often while writing, it is more likely to hurt than help, not because of a penalty, but because the text is harder to read and people leave faster.

The difference in one sentence

The idea that a term has to reach a certain frequency comes from a time when search engines counted words. That is long over.

Today's systems capture meaning. They recognize that having a website built, web design agency and getting a homepage made refer to the same field, without all three appearing in the text. They also recognize whether a text answers a question or writes around it.

What counts instead:

Completeness. Does the text answer the question and the obvious follow-up questions?

Recognizable expertise. Does it contain details only someone who has actually done the thing knows: numbers, processes, exceptions?

Structure. Can you skim it and still understand it?

The practical consequence: if a term appears unnaturally often while writing, it is more likely to hurt than help, not because of a penalty, but because the text is harder to read and people leave faster.

The answer belongs at the top

The most common structural mistake is the dramaturgy: introduction, lead-in, context, and the answer in the second-to-last paragraph.

That is the order of a talk, not of a text someone opens after a search. People who search want the answer immediately and decide within a few seconds whether to stay.

The better structure:

A heading that picks up the question. First paragraph: the answer in two to three sentences. Then the reasoning, the limitations, the examples.

This has two effects. People stay because they get something right away. And generative systems cite this paragraph because it contains a complete answer.

The worry that nobody keeps reading when the answer is at the top doesn't hold true. Anyone who gets the answer keeps reading to put it in context. Anyone who doesn't get it leaves.

Where the term belongs

There are four places where the main term should appear. That is all it takes.

The title: as far forward as possible, because the first words carry the most weight and are read first in search.

The H1: it can be phrased differently from the title but should contain the term.

The first hundred words: once, worked in naturally.

A subheading: if it fits the content, not by force.

Plus the URL, if the page is new. Changing an existing URL is almost never worth it.

Everything else follows while writing. Anyone writing about a topic automatically uses the related terms, plus variants, synonyms and technical terms that a system evaluates anyway.

A text in which the term appears thirty times is not an optimized text. It is a bad text about a good topic.

What makes a text concrete

The clearest difference between a text that ranks and one that sinks is the amount of concrete detail.

Vague: The costs vary depending on scope.
Concrete: Between 8,000 and 25,000 euros, depending on the number of pages and whether copy is included.

Vague: The process takes a few weeks.
Concrete: Six to ten weeks, two of them for the concept, three for design, two for the build, the rest for corrections.

Vague: There are various options.
Concrete: Three paths, each with the condition under which it fits.

Concrete detail has three effects: it proves experience, it answers the actual question, and it can be cited, by people and by systems.

The reason most texts stay vague is caution. A number can be wrong, a range can't. So: name ranges, add conditions and stick to them.

Structure you can skim

Very few people read an article from start to finish. Most skim it and read the sections that concern them.

What this means for the structure:

Subheadings that say something. What corporate design covers is a heading. Basics is not.

Short paragraphs. Three to five lines. A paragraph over twelve lines gets skipped on a smartphone.

Lists where things are enumerated. But not for everything. A text that consists only of lists has no line of thought.

Use emphasis sparingly. If every third sentence is bold, nothing stands out anymore.

One complete thought per section. That way, every section can be read and cited on its own.

The simple check: read only the headings. Does that make a sensible summary? Then the structure is right.

Length, freshness and the rest

On length. There is no right word count. There is a complete answer. Some questions are answered after 600 words, others need 2,500. A text artificially stretched to a target length contains filler, and that is noticeable.

As a guide: look at how long the first five results on your topic are. You don't have to be longer, but you shouldn't leave out anything important that is there.

On freshness. A date belongs visibly on the page and in the structured markup. For topics that change (prices, technology, law), an annual revision should be planned. An updated article is almost always more valuable than a new one on the same topic.

Finally, a point that is often missing: every article needs a next step. Not as an ad block, but as an offer: the matching service page, a related article, a conversation. A text without a next step creates knowledge, but no inquiry.