AI Search

Short vs. Long Content in AI Overviews: The Data Says Both Work

Despina Gavoyannis
Despina Gavoyannis is a Senior SEO Specialist at Ahrefs, a leading marketing platform for search, AI, and beyond. She has worked in SEO for over 10 years, specializing in revenue-driven strategies and collaborating closely with cross-functional teams, including UX designers, developers, and marketers.

Lately, I’ve seen some particularly bad advice floating around on how long content should be to show up in AI search. 

For instance, this study of only 20 brands (narrow sample size) is proclaiming that content needs 10,000+ words to be cited by AI.

Screenshot of a study on Reddit showing most-cited AI content has 10,000+ words, higher sentence counts, and better readability scores than low-cited content.

I’ve also seen others insist very short content (250–500 words) is the future for visibility in AI, arguing that AI models have limited context windows and prefer concise answers (they don’t).

In reality, it’s a chicken-or-egg situation. If most of the internet starts publishing 10,000+ word guides, that’s probably what will get cited. But they won’t get cited because they’re 10,000+ words. They’ll get cited because that’s what’s available (and what “fresh” content starts to look like).

But hey, don’t take my word for it.

In true Ahrefs fashion, we studied 174,000 pages cited in AI Overview responses to see what actually matters. Spoiler alert: there’s almost no correlation between word length and being cited in AI Overviews.

The best thing you can do is write as much as you need to convey your topic to your human audience concisely.

Trim the fluff, get straight to the point, and stop chasing arbitrary word counts “because it’s good for SEO” or “because AI prefers it.”

  • The average length of content cited in AI Overviews is 1282 words. This is slightly above the current average of content ranking in Google’s organic results, which is 1188 words.
  • There is a near-zero correlation between word length and being cited in AI Overview responses (0.04 Spearman correlation).
  • 53.4% of pages cited by AI Overviews are under 1,000 words.
  • 30.6% are between 1,000 and 2,000 words.
  • Only 16% are over 2,000 words.

The bottom line: content length isn’t a major factor in whether you get cited. Short pages (under 1,000 words), long-form guides (over 2,000 words), and everything in between can all work.

Our methodology
 We analyzed 560,346 AI Overviews and identified 1,677,876 cited URLs. After filtering for pages where we could successfully extract content, we ended up with 174,048 pages with valid data.

For each page, we stripped out HTML and boilerplate content, then measured the word count of the remaining raw text. We also tracked citation position (1–10) within each AI Overview to see whether content length affected where a page appeared in the citation list.

The average word count of pages cited in AI Overviews is 1,282 words. But averages can be misleading. Here’s how the distribution actually breaks down:

  • Under 350 words: 16.6%
  • 350–1,000 words: 36.8%
  • 1,000–2,000 words: 30.6%
  • Over 2,000 words: 16.0%

Word count correlation of pages in AI Overviews from Ahrefs' data study

More than half (53.4%) of all citations go to pages under 1,000 words. That’s traditionally considered “short” for blog posts and website content, and well below what most SEO professionals would request in a content brief.

Not exactly the 10,000+ word mega-guides some are recommending.

If longer content doesn’t get cited more often, maybe it at least ranks higher within AI Overviews? Nope.

The Spearman correlation between word count and citation position is 0.04 — essentially zero.

Here’s the average word count by position:

  • Position 1: 1,270 words
  • Position 2: 1,291 words
  • Position 3: 1,291 words
  • Positions 4–10: 1,690 words

Average word count by position in AI Overviews from Ahrefs' data study

There’s no meaningful difference in content length between the top three citation positions, though the correlation is so weak it’s barely worth considering.

When short content gets cited, its distribution across positions closely mirrors that of longer content.

For pages under 350 words:

  • 34% appear in position 1
  • 32% appear in position 2
  • 31% appear in position 3

Pages between 350 and 1,000 words show a similar pattern:

  • 30% in position 1
  • 32% in position 2
  • 34% in position 3

Over 95% of short content citations land in the top three positions. This suggests that when short content gets cited, it competes on equal footing with longer content for prominent placement in AI Overviews (and wins the battle).

Averages can be skewed by outliers. Medians give a better sense of what most cited pages actually look like.

Box plot showing word count distribution across 10 page types, with a median line at 1115 words. Audio pages show highest variance.

Here’s the median word count by page type:

Content FormatDescriptionMedian Word Count
ListingsEcommerce or marketplace listing pages.315 words
Core PagesFundamental website pages like Home, About and service landing pages.317 words
User-generated contentComments, posts and content generated by users on platforms like social media or Reddit.387 words
VideoVideo descriptions and transcripts where available.407 words
Interactive toolsContent within or surrounding online, interactive content, like calculators or free tools.507 words
Listing CollectionsEcommerce collection and product category pages.534 words
DocumentsPDFs, slides and similar documents in various file formats.676 words
ArticlesBlog posts and web articles.1166 words
AudioMainly podcast transcriptions.1226 words

Only Articles and Audio have a median higher than 1,000 words. (Audio pages are typically podcast transcriptions, which can run long. For instance, a 30+ minute conversation can be over 5,000 words.)

The median values of transactional and utility content (such as core pages, listings, and listing collection pages) typically range from 300 to 550 words and are still frequently cited. This matters for ecommerce and other non-blog content. You don’t need a 1,000+ word SEO-optimized buying guide to show up in AI Overviews. Match the format to the intent.

One more note: the overall median of 1,115 words is pulled up by the heavy representation of Articles and Audio files in our dataset. The longest blog post cited was around 3,500 words, and many audio files were between 3,000 and 5,500 words.

However, despite what some are suggesting, we’re not seeing 10,000+ word mega-guides dominating AI citations (at least not yet).

Content length alone won’t get you cited in AI Overviews. So what should you focus on instead?

  • Answer the query directly. Give people (and search systems) what they want early. Don’t bury the lede.
  • Prioritize structure and clarity. Use headings, lead with your main point, and write in declarative sentences that are easy to parse.
  • Write for humans first. No one is actually reading a 10,000+ word guide start to finish. If your content doesn’t get engagement, it won’t send the signals that influence search visibility in the first place.
  • Match length to the content type. A product page doesn’t need 2,000 words. A comprehensive guide might. Let the topic and format dictate the length, not an arbitrary word count target.

Both short-form and long-form content have their place. The goal is to write as much as you need to answer the question or cover the topic thoroughly. No more, no less.

Stop obsessing over word count and focus on actually answering the query instead.