The Short Version
- Most SEO content doesn’t rank because it’s a remix of whatever already ranks. Google has no reason to show a sixth version of the same article.
- Your content has two jobs. Match what the searcher wants exactly, then give them something the top results don’t.
- Search your keyword before you write anything. The results page tells you the type of page, the format and the angle Google wants.
- Put your keyword in the title, H1, URL and intro, then write like a person.
- Use AI for research and drafts if you want. The parts that make a page rank (experience, data, opinions) have to come from you.
Why Most SEO Content Doesn’t Rank
SEO content is anything you write to show up for a specific search, whether that’s a blog post, a guide, a category page or a landing page.
SEO writing is the process of making that content something Google wants to rank and people want to read. In practice, almost all of it fails.
Ahrefs looked at about 14 billion pages and found that 96.55% get zero traffic from Google. The three main reasons they gave are that nobody searches for the topic, the page has no backlinks, or the page doesn’t match what searchers actually want.
Two of those three are decisions you make before and while you write, which means they’re fully in your hands.
There’s another reason that’s harder to measure, and I think it’s the biggest one. Most SEO content is a lazy remix. The standard advice is to search your keyword, open the top 10 results, see what they cover and cover it too, ideally with a few hundred more words. Content tools make this even easier by scoring your draft on how closely it matches the pages already ranking. Everyone follows the same process, so everyone ends up writing the same article, and Google ends up with ten versions of one page.
So learning how to write SEO content that ranks comes down to two jobs. First, match the search intent exactly, so Google sees your page as a fit for the query. Second, add something the pages already ranking don’t have, so Google has a reason to pick yours over them.
Everything else (keywords, headings, title tags, links) is packaging around those two, and both of them start with the keyword you pick.
Pick a Keyword You Can Actually Win
A good keyword for a new piece of content passes three checks. People search for it, you can realistically compete for it, and ranking for it would actually do something for your business.
The first check is the one everyone obsesses over, and search volume matters less than it looks. A page that ranks well doesn’t rank for one keyword. Ahrefs’ 2017 study of 3 million searches found that the average #1 page also ranks in the top 10 for nearly 1,000 other keywords, with a median of around 400. So a keyword showing 300 searches a month can bring in a lot more than 300 visits, because the page picks up all the variations people type.
That’s also why you write one page per topic, not one page per keyword. “How to write SEO content” and “how to write content for SEO” are the same question asked two ways. Write two pages for them and you split your effort between two pages competing against each other. The quick check is to search both. If the same pages show up for both, it’s one topic and one page.
The second check is competition. Look at who’s ranking and how many different sites link to their pages. I walk through a ten-minute way to benchmark that in my guide on how to get backlinks. If every result is a huge brand with hundreds of sites linking to it, start with something smaller and more specific.
The third check is the one people skip. A post that ranks and brings in people who will never need what you sell is a nice number in Search Console and not much else. If you run an online store, my eCommerce keyword research guide goes much deeper on picking keywords that bring buyers.
Read the Search Results Before You Write Anything
Search intent is the reason behind a search, what the person actually wants to find when they type it.
You don’t have to guess it. The first page of results is Google’s current best answer to what people searching that query want, based on everything it knows about how they behave. So search your keyword, ideally in a private window so the results aren’t personalized, and study that page before you write a single word.
Ahrefs breaks this down into the three Cs of search intent, and it’s the most useful way I know to read a results page. I’d add a fourth thing to check, which is everything on the page that isn’t a regular result.
| What to check | What it tells you | Example |
| Content type | Whether Google wants a blog post, a product page, a category page, a tool or a video | “running shoes” is almost all product and category pages, so a blog post has no chance there |
| Content format | How the ranking pages are put together (a step-by-step guide, a list, a comparison, a review) | “best CRM for small business” is almost all list posts |
| Content angle | The main selling point the top pages share | “for beginners,” “in 2026,” “without experience,” “on a budget” |
| SERP features | What else people want from this search | People Also Ask questions, Reddit threads, videos, an AI Overview |
When I searched “how to write SEO content,” the first page was mostly how-to guides and tip lists from SEO tools and agencies, a couple of Reddit and LinkedIn discussions, and an AI Overview sitting on top. The People Also Ask box had questions like “Can ChatGPT write SEO content?” and “Is SEO dead now with AI?”
The content type there is a how-to guide. The format is a process with steps. And the questions show that the people searching this are worried about AI, so a guide that ignores it misses what’s on their mind.
Mismatched intent is one of the three reasons from that Ahrefs study, and it’s the easiest one to avoid. If Google shows ten comparison lists and you write a long essay, you won’t change its mind. You can, and should, do the same thing better. Don’t try to do something else.
Matching intent only gets you into the running, though. Every page already ranking matches it too.
Find What the Top Results Are Missing
This is the step that decides whether your page ranks.
Open the top five results for your keyword and read each one all the way through. Write down every point they all make.
That list is table stakes, and your page needs to cover it too, because it’s what Google has learned searchers expect. Then make a second list of what none of them do well. The questions they skip, the claims they make without any proof, the advice they all copy from each other, the parts that are out of date.
That second list is where your page wins.
There’s some evidence Google thinks about content this way. In 2022, Google was granted a patent called “Contextual estimation of link information gain,” which describes scoring a document on how much it adds “beyond information contained in documents that were previously viewed by the user.”
It also lines up neatly with Google’s helpful content questions, one of which asks whether your content provides “original information, reporting, research, or analysis.”
The something new usually comes from one of these places.
- First-hand experience. You’ve done the thing, used the product, run the test. Google asks directly whether your content shows first-hand expertise, “for example, expertise that comes from having actually used a product or service, or visiting a place.”
- Your own numbers. Prices from your last 20 projects, results from a test you ran, a quick survey of your customers. Small data is still data nobody else has.
- A real opinion. Most SEO content hedges everything until it says nothing. If you think one of the standard pieces of advice is wrong, say so and explain why.
- Better examples. Actual screenshots, actual before and afters, a worked example instead of “for example, a business might…”
- The question nobody answers. Check People Also Ask, Reddit threads on the topic, and the questions your customers ask you on calls and in emails. Those almost never make it into the top results, which is a glaring gap for anyone who does the work of collecting them.
- Being current. If the top pages cite a study from 2019 or recommend a tool that no longer exists, the up-to-date version is a real improvement.
When I read the top-ranking guides for this keyword, nearly all of them covered the same checklist in roughly the same order. Pick a keyword, check the intent, use headings, write a title and meta description, add internal links.
Only one of the six I read spent real time on making the content different from what already ranks, even though that’s the part that decides whether you rank at all. That gap is the reason this section exists.
Content optimization tools like Surfer and Clearscope can work against you here. They’re useful for spotting a subtopic you forgot. But they build their recommendations from the pages already ranking, so a perfect score mostly means you’ve written a very good average of the top 10.
But the average already exists. Google doesn’t need another one.
Outline Around the Answer
Once you know what to cover and what you’re adding, sketch the structure before you write a full draft. Most of an article’s readability gets decided here, long before the sentences.
Put the answer first, in the article as a whole and in every section. Journalists call this the inverted pyramid, it’s the right format for the web because people skim and stop reading early.
The idea is that the most important information “(or what might even be considered the conclusion) is presented first.” If someone reads only the first two sentences under a heading, they should still get the point of that section.
For headings, use the questions the reader actually has, in the order they’d ask them.
The People Also Ask box, Google’s autocomplete and your table stakes list from the last step give you most of them. A heading like “Why Most SEO Content Doesn’t Rank” tells the reader what they’re about to get. A heading like “The Importance of Quality” tells them nothing.
Don’t set a word count though. There’s no magical word count target, minimum or maximum, and ask yourself whether you’re writing to a particular word count because you’ve heard or read that Google has a preferred word count.
The competing pages give you a rough idea of how deep the topic goes. After that, the right length is however long it takes to answer the question fully, and not a paragraph more.
Put whatever you’re adding early in the piece, too. If your data or your take is what makes the page worth reading, don’t make people scroll through six standard sections to find it.
Write It Like You’re Explaining It to One Person
Most people skim. Jakob Nielsen’s analysis of real page views found that on the average page, users have time to read at most 28% of the words, and “20% is more likely.”
That study is from 2008, and I don’t think people’s attention spans have improved since.
So write for the skimmer first. Clear headings, the answer at the top of each section, a table when you’re comparing things and a list when there really are separate items.
Then write for the person who does read everything, in plain words, like you’re explaining it to one person sitting across the table from you.
Cut the warm-up, because nobody needs three sentences on how important SEO is before you tell them how to do it. Use specifics instead of generalities (“reply the same day” instead of “be responsive”).
Explain a term the first time you use it, briefly, in the same sentence. And when you know something because you’ve done it, say so, because that’s exactly the first-hand experience Google’s guidelines keep asking for.
Google frames this as asking “Who, How, and Why” about your content. Who wrote it, and is that obvious from a byline that leads to some background on the author?
How was it made, and if AI or automation did a lot of the work, is that clear to readers?
And why does it exist, to help people or mainly “to attract search engine visits”?
Google is also clear in its starter guide that E-E-A-T (experience, expertise, authoritativeness and trust) isn’t a ranking factor in itself. The things it describes are what Google’s systems try to reward, though, and of the four, Google says “trust is most important.”
Where Keywords Go (and Where They Don’t)
Put your main keyword in the title tag, the H1, the URL and somewhere in the first paragraph or two. Use it, or a close variation, in a subheading or two where it reads naturally. That’s about it.
After that, write the way people actually talk about the topic.Most of the other keyword placement advice you’ll find is myth.
As for how many keywords to use per page, think in topics instead. One page, one main topic, and the related keywords come with it. That’s how pages end up ranking for hundreds of searches they never targeted directly.
Get the On-Page Basics Right
On-page SEO covers everything on the page itself that helps Google understand it and helps people decide to click it.
None of it will make a weak article rank, but getting it wrong can hold back a good one, and it takes about ten minutes.
| Element | What to do | Where people go wrong |
| Title tag | Unique, clear, accurate, main keyword near the start | Stuffing in keywords or making it vague |
| Meta description | A short, specific summary that makes people want to click | Treating it as a ranking factor, or leaving it blank |
| URL slug | Short and readable, like /how-to-write-seo-content/ | Dates, filler words, long auto-generated slugs |
| H1 and subheadings | One H1 that matches the title, H2s for sections, H3s under them | Using headings to make text bigger instead of to structure the page |
| Images | Descriptive file names and alt text that describes the image | Keyword-stuffed alt text, or none at all |
| Internal links | Links to the new page from related pages, with descriptive anchor text | Publishing it with nothing linking to it |
| External links | Links to the sources behind your claims | Stats with no source |
Google rewrites a lot of title tags. In a 2025 study, John McAlpin found Google changed 76% of the titles in his dataset, but the most common change (63% of rewrites) was simply dropping the brand name.
Google’s documentation on title links lists the usual reasons for rewrites, like titles that are half-empty, out of date, inaccurate, repeated across a bunch of pages, or repeating the site name. A clear, accurate title gets rewritten less.
Meta descriptions get rewritten almost as often, and they don’t affect rankings at all. However, write one anyway, because when Google does show it, it’s your ad on the results page.
How to Use AI Without Writing the Same Article as Everyone Else
Google doesn’t mind AI. What it cares about is whether the result is any good, and whether you’re using it to flood search results with pages.
Its guidance on AI content from February 2023 says “appropriate use of AI or automation is not against our guidelines,” while using it “to generate content with the primary purpose of manipulating ranking in search results is a violation of our spam policies.”
In March 2024, Google went further and expanded its spam policy on scaled content abuse to cover mass-produced pages “whether automation, humans or a combination are involved.” By late April, Google said search results had 45% less low-quality, unoriginal content (a number Google measured itself, so take the exact figure with a pinch of salt).
The data on how AI content ranks looks contradictory at first. Ahrefs analyzed 600,000 top-ranking pages and found 86.5% contain some AI content, with basically zero correlation between how much AI a page used and where it ranked. Graphite ran its own study and found 86% of ranking articles were written by humans.
The two studies are measuring different things. Ahrefs counted any AI in the page, while Graphite classified whole articles as either AI-generated or human-written. Put together, AI help is completely normal on pages that rank, and fully AI-generated articles will inevitably struggle.
In Graphite’s data, AI articles made up only 7% of #1 results, about half of what you’d expect from their share of results overall. Both studies rely on AI detectors, which get things very wrong, so I wouldn’t lean on the exact numbers.
You should also know that a language model writes the most likely version of an article on a topic, which is more or less the average of everything already written about it.
That’s the remix problem from the start of this guide, just automated and a lot faster. It also can’t add the things that make content worth ranking, because it hasn’t run your test, doesn’t have your numbers and has never talked to your customers. Left alone, it produces the soulless, perfectly structured article that already exists ten times over.
However, you should use it where it’s good. Summarizing research, drafting a first outline, spotting what the top results cover, writing a rough version of the standard sections, tightening your sentences.
Keep the parts it can’t do (the experience, the data, the opinions, the examples) and the final edit for yourself. And check every number and every claim, because AI will confidently cite a study that doesn’t exist.
Publishing Is Where the Work Starts
Rankings take time, and most new pages never get there. Ahrefs found that only 1.74% of newly published pages reach the top 10 within a year, and 72.9% of pages in the top 10 are more than three years old. The pages that do make it tend to move early, though. About 41% of them got into the top 10 within the first month. So don’t panic in week two, but if a page has gone six months without ranking, it needs work.
When a good page doesn’t rank, the reason usually sits outside the writing.
The first thing to check is links. If the pages above you have 40 sites linking to them and yours has none, better writing won’t close that gap, and my guide on backlinks that move rankings covers how to fix it.
The quicker fix is internal links. Go through your existing pages on related topics and link them to the new one, so Google finds it sooner and your stronger pages pass some of their value to it.
After a few months, update the page using real data instead of guesses. Open the Performance report in Google Search Console, filter to the page, and look at the queries it’s getting impressions for.
You’ll usually find searches you’re ranking on page two for without fully answering them. Add a section for those, beef up the parts that are thin, and check whether the results page itself has changed since you wrote the piece.
What you shouldn’t do is change the publish date and call it an update.
Is SEO Content Still Worth Writing With AI Search?
Yes, but the bar is higher than it was two years ago. AI Overviews answer a lot of simple questions right on the results page, and fewer people click through. Pew Research tracked the Google searches of 900 US adults in March 2025 and found they clicked a regular result in 8% of visits that had an AI summary, compared to 15% without one. They clicked a link inside the summary itself in just 1% of visits.
Which content loses depends on what it offers. The pages hit hardest are the ones that only restate the basics, because restating the basics is exactly what an AI summary does well.
A definition of SEO content doesn’t need a click anymore. Your test results, your data, your worked examples and your opinion still do, because the summary can only point to them.
Ranking isn’t the only way into those answers anymore, either. Only 38% of AI Overview citations come from pages in the top 10, down from 76% in an earlier study, which they put down to Google pulling sources from a wider set of related searches.
I don’t think anyone knows yet exactly how to optimize for that, and I’d be careful with anyone who says they do. What the cited pages do have in common is that they answer a specific question clearly and have something in them worth citing, which is the same thing that gets a page ranked in the first place.