When worse writing outranks you, the problem isn't the writing

Somewhere in your market there is a page that outranks yours and reads like nobody cared. It is thin, it repeats itself, and it answers the question worse than you do. The instinct is to conclude that search is broken, that quality no longer matters, or — increasingly — that Google is quietly holding your pages down for how they were produced. All three conclusions are wrong, and acting on any of them costs you a quarter.

Google ranks pages, not reputations

The diagnosis usually begins as a comparison between two companies. Our site is better than theirs, our team is better than theirs, therefore our page should win. That is not the comparison the ranking systems make.

Google's guide to its ranking systems says the systems are "designed to work on the page level", with site-wide signals contributing to the understanding of a page rather than deciding it. The same page adds the part people skip: poor site-wide signals do not mean all of a site's content ranks poorly, and good ones do not guarantee that any of it ranks well.

So the competitor's dated design, their neglected homepage, their three-sentence About page — none of it is evidence in your favour. The contest is one page against one page, for one query. Narrow the question that far and the mystery usually shrinks into something specific enough to act on.

If your first hypothesis was that machine-written pages are being suppressed, that is the wrong hypothesis, and we have set out the reasoning at length in Google doesn't penalize AI content. Authorship is not the variable here.

Links and time are doing the work you credited to prose

On that same ranking-systems page, Google names link analysis and PageRank among its core systems. This is not a secret lever inferred from forum gossip; it is documented, first-party, and decades old. A page that has sat in the market for four years, collecting citations from people writing about the topic, has accumulated something your three-month-old post has not. No amount of editing closes that gap directly, because the gap is not on the page.

Age itself is not the ranking factor folklore imagines. What age buys is exposure: time to be discovered, quoted, linked and updated. It also buys evidence. A page that has satisfied this query for years is a lower-risk answer than one that has never been tested against it.

Here is the uncomfortable version, from a company that sells writing. Prose quality is a threshold, not a slope. Below the threshold — inaccurate, unclear, not actually answering the question — nothing else you do will help. Above it, further polish buys remarkably little. The page you are staring at cleared the threshold. It just did it four years ago.

The query may not be asking for what you wrote

Read the results page instead of the competitor. If the top five results are specification tables, comparison grids and prices, and yours is a considered twelve-hundred-word essay, the query is asking for a different shape of answer. The better essay loses to the worse table every time, because searchers keep choosing the table.

Freshness is the other half of it. Google documents "query deserves freshness" systems that surface recent content where recency is expected. Some queries are permanently tilted toward whatever was published this quarter. You cannot out-write that tilt. You can only publish into it again.

Rewriting the losing page is the least likely fix

The reflex is to rewrite. Google's guidance on debugging drops in Search traffic is unusually blunt about those odds: there is no guarantee that changes to your site produce noticeable impact, and if there is more deserving content, it will continue to rank well. The same page advises against making radical changes to a page that is already performing.

Timescale makes it worse. That documentation notes that some changes take effect in days and others in months, and that it can take months for Google's systems to determine a site is producing helpful content over the long term. So a rewrite costs an afternoon and returns its verdict a season later, hopelessly entangled with everything else that happened in between. As a way of learning what works, it is close to useless.

What we would change instead

Publish the adjacent coverage rather than re-polishing the one page. Depth across a topic compounds and can be scheduled; polish on a single URL does neither. Then make the archive link to itself properly, so the authority you already have is not stranded on pages nobody reaches — the cheapest SEO hour you are not spending. Then give it two quarters before you judge any of it.

And the honest limit, since this is the point where automation stops being the answer. An engine can produce the twelve adjacent posts on schedule, in both your languages, and keep the internal links current while it does. It cannot make anyone link to you. The thing actually separating you from that worse-written page — having been talked about — is bought with a real result, a real dataset, or a real opinion somebody in your industry wants to argue with. A machine can publish the argument. It cannot supply the conviction behind it.

That competitor is not beating you at writing. They are beating you at having been there longer and at being cited more often, and neither of those is answered by another draft of the page that lost.