Does AI content perform well in search? [Survey + Data study]

Using AI in content production has become a common practice. A question many SEO teams ask is how well it works.

So we analyzed 20,000 keywords and 42,000 blog posts to find out.

Content categorized as AI-generated appeared at the top just 9% of the time. Content categorized as written by humans was present 80% of the time.

But that number needs context. The finding is not that AI is inherently bad; is that the search rewards one’s personality. The detector doesn’t care about your process. Reads the finished product.

This study looks at how marketers perceive the impact of AI on content performance, and how those perceptions hold up to real-world ranking data.

Important takeaways

• 87% of SEO teams report that their content is entirely human-created or largely human-led, keeping people directly involved in production and planning.
• 72% of SEOs say that AI content ranks at least as well as human-written content. But our analysis of 42,000 blog pages shows that 1st position results are 8x more likely to be written by a human.
• 70% of SEO teams cite speed as the top benefit of using AI, but only 19% say it improves content quality.
• The use of AI is concentrated in text-based tasks but drops significantly in multimedia and localization, where the tools are more specialized and the output harder to evaluate.
• Human-authored content is profitable in all of the top 10 SERP positions, but the gap is sharpest near the top where AI-generated content almost doubles between position 1 and position 4.

1. Humans still lead most content apps

87% of SEO teams report that their content is entirely human-created or mostly human-led. Keeping people directly involved in production and planning.

64% of SEOs use human-led AI workflows, making it the most common content generation model today.

Another 23% report creating content entirely without AI.

This suggests that while AI is becoming a common infrastructure in content workflows, many teams have people who continue to play a key role in the planning process.

2. AI is often used in content creation for tasks such as research, planning, and optimization

At least 65% of respondents use AI for core writing tasks such as research, editing, and on-page optimization.

But the use of AI is very low in tasks that go beyond the writing process. Visual content creation (28%), translation (15%), and video or audio production (9%) are far behind.

That gap makes sense.

Text-based tasks are where AI feels most natural and easy to integrate. However, multimedia and localization require special tools, specific judgment, and a lot of budget.

Working on the Semrush blog, AI supports the way we generate content, but it only gets part of the way. At every step in which we use AI — ideation, specification, writing… — we have someone reviewing the output. Every essay involves at least three people (strategist, writer, and editor). Usually up to five. And they are all subject matter experts or work very closely with them.

They also have a deep understanding of our tools and how to use them to address the real problems our customers face. That layer is critical. In SaaS, especially for a product as broad as Semrush, recommending the right solution requires context and judgment.

And – at least in our experience – AI tools never do a very good job, no matter how much information we feed them. Another area where we draw a clear line today is planning. That initiative is still fully led by the people. We experimented with internal tools to support fact-checking. But we haven’t changed the programming decision to include AI, and we’re in no rush to do so.

Ana Camarena, Head of Organic Content Strategy at Semrush

3. Faster help and insights are the biggest advantages of AI

70% of SEO teams say faster content generation is the top benefit of using AI. Help with thinking and thinking comes in second place with 62%.

But only 19% say AI improves content quality.

That space is telling.

It may be because quality is difficult to define neatly. It’s usually a case of knowing it when you see it.

And if you can’t clearly define what quality looks like, it’s hard to give AI tools clear direction to deliver it.

4. Many SEOs think that AI content and human written content

The majority of SEO professionals who use AI content (72%) say they think AI-assisted content performs as well or better than human-written content in search rankings—that’s up from 64% in our 2024 survey.

The share that they say is doing the worst has grown, from 9% to 13%.

The biggest change is in the “not sure / haven’t compared” category, which has dropped from 27% to 15%.

This shows how ideas flow in both directions. After having time to try and implement AI in their workflow, many teams now have a clear idea of ​​how they think AI content works.

5. SEOs report AI content performance has improved

About 45% of respondents who use AI content say their SEO performance has improved over the past year. Another 25% said that the performance is always the same.

Only 6% reported a decline.

But 25% say it’s too early to tell or they haven’t tracked performance yet.

It’s possible that those groups don’t clearly differentiate between their human-written and AI-assisted content.

Or they may not have a clue in place to measure the difference.

That’s noteworthy, because without a clear definition, it’s hard to know whether performance gains come from AI, human programming, or a combination of the two.

6. Our analysis of 42,000 blog posts shows that pages written by people still win

After grading 42,000 blog posts with an AI detector, content classified as full human-authored content was classified as AI-generated or mixed in all of the top ten positions.

But the gap is most striking in position 1, where pages have an 80.5% chance of being written by a human compared to only 10% generated by AI.

This may seem contradictory to survey data, where 72% of SEOs say AI content ranks at least as well as human-written content.

But that’s not the case.

The only real difference is much higher.

From 5th place onwards, the gap between human written content and AI production is very small.

So when multiple teams benchmark against “rank on the first page,” AI content holds its own. It’s only when you zoom in that the human-written content clearly pulls you forward.

What does this research mean for marketing teams

AI is now a common part of content production for many teams. However, it is not a hands-free shortcut to ranking higher in search.

For SEO teams, the implication is straightforward:

Use AI to move faster through research, presentation, and writing. Then invest the time you’ve saved in incorporating expert insights, proprietary data, and other elements that make your content stand out.

AI helps us move faster, and speed matters. But it’s not enough to justify lowering quality standards. In most cases, an extra day or two will not change the result – the quality will change. AI can support more efficient content. But the content that stands out is still shaped by strong human input. And that often means trading some speed for better judgment and deeper expertise.

Ana Camarena, Head of Organic Content Strategy at Semrush

Want to use Semrush’s AI content generator?

Try the Content toolkit.

Instead of starting with a blank page, you start with a brief statement that already shows what’s standard – and what’s not. Use it to quickly move between first drafts and add your own personal details, thoughts, and ideas to elevate the content you’ve produced to a rankable level.

The perfect way

We wanted real answers, so we dug into the data from two sides.

Here’s what we did:

1. 42,000 blog page rank analysis: In November 2025, we collected 20,000 keywords, pulled out the top 10 Google results for each (200,000 URLs), and filtered only blog pages (URLs containing “/blog/”) — leaving us with 42,000 pages. We split each article into pure text and ran it through GPTZero to classify it as human-written, AI-generated, or hybrid. We then calculate the probability of each type of content appearing in all SERP positions.
2. A survey of 224 SEO experts: We surveyed 224 marketers working hard on SEO and content marketing. The majority of respondents worked directly with content: 58% reported that they create and develop content, and 40% said that they can curate or approve content created by others. All results represent percentages of the total sample.