Short-form vs long-form content statistics at a glance
| Category | Statistic | Source |
|---|---|---|
| Rankings | The average Google top 10 result runs 1,447 words, with no direct relationship between word count and position, across 11.8 million results | Backlinko, April 2025 |
| Rankings | Google states it has no preferred word count for ranking | Google Search Central, 2026 |
| Links | Content above 3,000 words earns 77.2% more referring domains than content below 1,000 words, across 912 million posts | Backlinko and BuzzSumo, 2019 |
| Links | Referring domains per 1,000 words fall about 62% when you move from a 750-word post to a 3,500-word post | Computed from Backlinko and BuzzSumo, 2019 |
| Shares | Articles of 1,000 to 2,000 words earn 56.1% more social shares than articles below 1,000 words | Backlinko and BuzzSumo, 2019 |
| Production | The average blog post is 1,312 words in 2026, down from a peak of 1,427 in 2023 | Orbit Media, 2026 |
| Production | The average post takes 3 hours 20 minutes to write, down from 4 hours 10 minutes in 2022 | Orbit Media, 2026 |
| Attention | Readers have time to read at most 28% of the words on an average visit, and 20% is more likely | Nielsen Norman Group, 2008 |
| Attention | 55% of visits spend fewer than 15 seconds actively on a page, across 2 billion visits | Chartbeat via Time, 2014 |
| Video | Videos under one minute average a 52% engagement rate, across more than 13 million videos | Wistia, 2026 |
| Video | 65% of viewers stay engaged to the end of a video under one minute; 20% do on videos over 20 minutes | Vidyard, 2025 |
| Video | A 20-minute video delivers about 6.2 times the watched minutes of a one-minute video | Computed from Vidyard, 2025 |
| AI search | 53.4% of pages cited in Google AI Overviews are under 1,000 words; 16.6% are under 350 | Ahrefs, December 2025 |
| AI search | The correlation between word count and AI Overview citation is 0.04, near zero | Ahrefs, December 2025 |
Does longer content rank better in Google?
Longer content does not rank better in Google. Backlinko's 2025 study of 11.8 million results puts the average top 10 page at 1,447 words and found no direct relationship between word count and ranking position, so long pages fill page one without length being the reason.
| Statistic | Source |
|---|---|
| The average word count of a Google top 10 result is 1,447 words. Backlinko reports no direct relationship between word count and rankings, and word count across the top 10 is evenly distributed. | Backlinko, 11.8 million results, April 2025 |
| The average word count of a current Google organic result is 1,188 words, measured on raw text with HTML and boilerplate stripped. | Ahrefs, 174,048 pages, December 2025 |
| Google's own guidance asks writers whether they are targeting a word count because they heard Google prefers one, and answers "No, we don't." | Google Search Central, 2026 |
| The #1 result in Google has on average 3.8 times more backlinks than the results in positions 2 to 10. | Backlinko, April 2025 |
| Roughly 95% of all pages have zero backlinks. | Backlinko, April 2025 |
Backlinko says plainly that this is a correlation study and that it cannot pinpoint why long-form content tends to appear on page one. The average published post is 1,312 words (Orbit Media, 2026) against 1,447 words for the average top 10 page (Backlinko, 2025). Those two averages come from different years and different populations, so the 135-word gap is a rough comparison, and it is the whole of the ranking case for length.
I have watched teams pad a 900-word answer to 2,000 because a tool told them the top 10 averaged 2,000. Backlinko and Ahrefs both say the same thing out loud: length travels with rankings, it does not cause them. Length is a proxy for having covered the topic and for having earned links, and if you add words without adding either, you have bought nothing. The link data is where the length effect really lives.
How long is the average blog post in 2026, and what does it cost to write?
The average blog post is 1,312 words in 2026 and takes 3 hours 20 minutes to write, according to Orbit Media's survey of 1,042 content marketers. Both numbers have been falling since 2022.
| Statistic | Source |
|---|---|
| The average blog post is 1,312 words in 2026, down from a peak of 1,427 words in 2023. | Orbit Media, 1,042 content marketers, 2026 |
| The average time spent on a post is 3 hours 20 minutes in 2026, down from a peak of 4 hours 10 minutes in 2022. | Orbit Media, 2026 |
| 13.9% of content marketers report strong results in 2026, below the 20% to 30% range the survey recorded in earlier years. | Orbit Media, 2026 |
| 92.4% of content marketers now use AI for blogging, and AI adoption shows no correlation with better results in the survey. | Orbit Media, 2026 |
| Recomputed: at 200 minutes per 1,312-word post, 1,000 words of copy costs about 152 minutes of writing time. | Computed from Orbit Media, 2026 |
Post length has fallen for three years in a row and the share of marketers reporting strong results fell with it, which is the awkward pairing nobody in this debate likes.
Everyone reads the 1,312 words as proof that short won. Read the other column. Effort per post dropped 50 minutes since 2022, AI adoption hit 92.4%, and reported results fell to 13.9%. Shorter posts did not cause that; cheaper posts did. Length is the wrong dial, and what Orbit measures as effort, meaning research, original data and editing, is the one that still moves.
How much of a long article do readers read?
Readers get through at most 28% of the words on an average visit, according to Nielsen Norman Group's 2008 analysis of 45,237 page views. Each extra 100 words earns only 4.4 more seconds of their time.
| Statistic | Source |
|---|---|
| Users have time to read at most 28% of the words during an average visit, and 20% is more likely. DATED, published May 2008, still the most precise measurement of the effect. | Nielsen Norman Group, 45,237 page views, 2008 |
| Users spend only 4.4 seconds more for each additional 100 words on a page, against a baseline of roughly 25 seconds. DATED, 2008. | Nielsen Norman Group, 2008 |
| 55% of visits spend fewer than 15 seconds actively on a page, across 2 billion visits, 580,000 articles and 2,000 sites. DATED, 2014. | Chartbeat via Time, 2014 |
| Engagement across the web, measured as page views, scroll rate and time per visit, fell 10% year over year in Q4 2025 against Q4 2024, across 6,500+ websites and 99 billion sessions. | Contentsquare Digital Experience Benchmark, 2026 |
| Recomputed: 1,000 extra words buys about 44 seconds of extra reading, against about 152 minutes of writing if writing time scales with length, a ratio of roughly 208 to 1. | Computed from Nielsen Norman Group 2008 and Orbit Media 2026 |
The Nielsen Norman numbers are eighteen years old and flagged as such in the table, because nothing published since has replaced the measurement. Contentsquare's 2026 benchmark points the same way on a far larger sample: engagement per visit is falling.
The 4.4 seconds figure is the most useful number on this page and the least cited. Reading time does grow with length, and it grows so slowly that length is nearly free for the reader and expensive for you. That asymmetry is why long form works when the reader came to solve something specific and fails when you are hoping to hold general interest. Video runs the same pattern with very different arithmetic.
Does short-form or long-form video hold attention better?
Short-form video holds a higher share of its audience and long-form video delivers more total watched minutes. Vidyard's 2025 benchmark found 65% of viewers stay to the end of a video under one minute, against 20% on videos over 20 minutes. That works out as roughly 6.2 times more watched minutes for the 20-minute video.
| Statistic | Source |
|---|---|
| Videos under a minute average a 52% engagement rate, across more than 13 million videos uploaded to Wistia. | Wistia State of Video, 2026 |
| Videos between five and 30 minutes average a 9% click-through rate, and engagement drops after the five-minute mark. | Wistia, 2026 |
| Engagement falls a further 11% once a video crosses 30 minutes. | Wistia, 2026 |
| 65% of viewers stay engaged to the end of videos under one minute; 20% do on videos over 20 minutes, across 943,305 videos created during 2024. | Vidyard Business Video Benchmarks, 2025 |
| Videos over 20 minutes grew 420% year over year in creation volume, against about 30% for videos under three minutes and 60% for three to 20 minutes. | Vidyard, 2025 |
| TikTok users spend 1 hour 37 minutes a day in the Android app, 14% longer than YouTube users spend in theirs. August 2025 data. | DataReportal Digital 2026, October 2025 |
| Online video is the highest-reaching weekly media activity, used by 91.1% of adult internet users each week. | DataReportal, 2025 |
| Recomputed: one minute at 65% retention delivers 0.65 watched minutes; 20 minutes at 20% delivers 4.0, a 6.2x gap in the long video's favour. | Computed from Vidyard, 2025 |
Retention rate and watched minutes are different metrics and the short-form case usually quotes only the first.
If you measure video in percentages, short form wins every chart. Measure it in minutes of human attention and a 20-minute video beats a one-minute video by six to one on the same retention curve. Both are true and they answer different questions. Use short form to get found and long form to get remembered, and stop treating completion rate as a quality score.
Are people moving to short-form video for information?
People are moving to short-form video for news faster than to any text format. The Reuters Institute's 2026 report finds 77% of people worldwide watch online news video weekly. Pew Research Center puts 20% of US adults on TikTok for news in 2025.
| Statistic | Source |
|---|---|
| 77% of people globally consume online news video each week, and for the first time a majority do so in all 48 markets surveyed. Sample: 97,520 respondents, fieldwork in early 2026. | Reuters Institute Digital News Report, 2026 |
| 27% of respondents get news from news-focused creators and 46% from creators of any type, while podcast use is flat at 11%. | Reuters Institute, 2026 |
| 20% of US adults regularly get news on TikTok, and 55% of TikTok users do, up from 22% of users in 2020. Survey fielded 18 to 24 August 2025. | Pew Research Center, 2025 |
| TikTok averages a 2.60% engagement rate by followers against 0.48% on Instagram, 0.15% on Facebook and 0.12% on X, across 70 million posts from January 2024 to December 2025. | Socialinsider, 2026 |
| Pictures earn 35% more engagement than text posts on Facebook and nearly 44% more than videos, across 52 million Facebook posts. On X, text-only posts beat videos, images and links, and PDF carousels hold the highest median engagement on LinkedIn across more than 2 million posts. | Buffer, January 2026 |
Short video wins the information habit and loses on several of the platforms Buffer measured, which tells you the format question is really a platform question.
The Reuters Institute number should change your plan: a majority now watch news video in all 48 markets, and that is a habit, not a trend. What it does not say is that text lost. Buffer has text-only posts beating video on X and PDF carousels topping LinkedIn on the same day TikTok posts a 2.60% engagement rate. Pick the format the platform rewards instead of porting one format everywhere.
What content length do AI search engines cite?
AI search engines cite short pages more often than long ones. Ahrefs' December 2025 study of 560,346 AI Overviews found 53.4% of cited pages run under 1,000 words. The correlation between word count and citation is 0.04.
| Statistic | Source |
|---|---|
| 53.4% of URLs cited in AI Overviews are pages under 1,000 words, and 16.6% are under 350 words. Sample: 560,346 AI Overviews, 1,677,876 cited URLs, 174,048 pages with valid word counts. | Ahrefs, December 2025 |
| The Spearman correlation between a page's word count and whether it is cited in an AI Overview is 0.04, and the correlation with citation position is also 0.04. | Ahrefs, December 2025 |
| Pages cited in AI Overviews average 1,282 words against 1,188 words for current Google organic results, a difference of 94 words. | Ahrefs, December 2025 |
| Citation positions 1 to 3 average 1,271 to 1,291 words while positions 4 to 10 average 1,690, so shorter pages sit higher in the citation list. | Ahrefs, December 2025 |
| Median word counts among cited pages: listings 315, core pages 317, video pages 407, articles 1,166. Medians, not averages. | Ahrefs, December 2025 |
| Keywords that trigger AI Overviews tend to be longer and more specific, and about 60% of them have 100 or fewer monthly searches, across more than 10 million keywords. | Semrush, 2025 |
The distribution is the finding, not the average. A third of AI Overview citations go to pages between 1,000 and 2,000 words and a sixth go to pages under 350, so both ends of the range get quoted.
A 0.04 correlation means length is not a lever for AI citation at all. What gets you quoted is a passage that answers the question on its own, and a 300-word listing page can carry one as well as a 3,000-word guide. If you are rewriting your library for AI search, break long pages into self-contained answers rather than cutting words. The same logic drives content decay once a page stops matching the question people ask.
How we calculated the original numbers
Four figures here do not appear anywhere else. Here is the arithmetic, so you can check it or change the inputs.
- 208 to 1: minutes of writing per extra minute of reading. Orbit Media reports an average 1,312-word post written in 200 minutes, so 1,000 words costs about 152 minutes (200 / 1,312 x 1,000 = 152.4). Nielsen Norman Group measured 4.4 extra seconds per extra 100 words, so 1,000 extra words buys 44 seconds, or 0.733 minutes. 152.4 / 0.733 = 207.9. This assumes writing time scales with length, which is generous to long form.
- 6.2x: watched minutes, 20-minute video against one-minute video. Vidyard's 2025 benchmark of 943,305 videos created during 2024 reports 65% of viewers engaged to the end below one minute and 20% above 20 minutes. One minute x 0.65 = 0.65 watched minutes; 20 minutes x 0.20 = 4.0. 4.0 / 0.65 = 6.15. End-of-video retention stands in for average completion, which Vidyard does not publish.
- 62%: the fall in referring domains per 1,000 words. Backlinko and BuzzSumo found content above 3,000 words earns 77.2% more referring domains than content below 1,000. Taking 750 and 3,500 words as representative, the long piece carries 4.67 times the words for 1.772 times the links. 1.772 / 4.67 = 0.379, so link yield per 1,000 words falls 62.1%. Other bracket midpoints move the number but never above parity.
- 135 words: the gap between what marketers publish and what ranks. Backlinko's 2025 study puts the average Google top 10 page at 1,447 words and Orbit Media's 2026 survey puts the average published post at 1,312. 1,447 - 1,312 = 135, or 9.3%. Both are averages rather than medians, and they come from different years and different populations, ranked results against self-reported publishing. Treat the gap as an order of magnitude, not a target.
What the 2026 numbers say
Length is a symptom, not a lever. Backlinko found no direct relationship between word count and rankings across 11.8 million results. Ahrefs found a 0.04 correlation with AI Overview citations across 174,048 pages, and Google's own documentation says it has no preferred word count. Three measurements, three methods, and none of them puts length on the causal side.
The per-word economics favour short, the per-piece economics favour long. Long posts pull 77.2% more referring domains and 62% fewer per 1,000 words. Long videos lose two thirds of their retention rate and gain six times the watched minutes. Which one matters depends on whether your constraint is production capacity or audience attention, and most teams have never written that down.
Format follows platform, not doctrine. Short video owns the information habit, with 77% weekly news video use worldwide and TikTok at a 2.60% engagement rate, while text-only posts still beat video on X. The teams that win 2027 will run a 300-word answer page, a 3,000-word reference and a 45-second clip on the same topic, each measured against a different target. The content marketing ROI numbers put a price on that, and the rest of the series sits on the statistics hub.
FAQ
Is long-form or short-form content better for SEO?
Neither, based on the two biggest correlation studies. Backlinko analysed 11.8 million Google results in April 2025 and found no direct relationship between word count and ranking position, although the average top 10 page runs 1,447 words. Google's own documentation states it has no preferred word count.
How long should a blog post be in 2026?
The average blog post is 1,312 words in 2026, according to Orbit Media's survey of 1,042 content marketers, down from a 1,427-word peak in 2023. The average Google top 10 page is 1,447 words (Backlinko, 2025). That 135-word gap sits between two different studies, years and populations, and it is small enough to ignore in favour of search intent.
Do longer articles get more backlinks?
Yes in total and no per word. Content above 3,000 words earns 77.2% more referring domains than content below 1,000 words, across 912 million posts (Backlinko and BuzzSumo, 2019). That is 1.77 times the links for about 4.7 times the words, so referring domains per 1,000 words fall about 62%.
What is the ideal video length for engagement?
Videos under one minute average a 52% engagement rate across more than 13 million videos (Wistia, 2026). Vidyard found 65% of viewers finish a video under a minute against 20% on videos over 20 minutes, across 943,305 videos in 2025. Measured in watched minutes rather than percentages, the 20-minute video delivers about 6.2 times more attention.
Does AI search prefer short or long content?
Short, slightly. 53.4% of pages cited in Google AI Overviews are under 1,000 words and 16.6% are under 350, across 560,346 AI Overviews (Ahrefs, December 2025). The correlation between word count and citation is 0.04, effectively zero.
How much of an article do people read?
At most 28% of the words on an average visit, from Nielsen Norman Group's analysis of 45,237 page views, and each extra 100 words earns only 4.4 seconds of extra reading. Chartbeat's analysis of 2 billion visits found 55% of visits last under 15 seconds of active time. Both measurements are dated and no larger study has replaced them.
Sources
- Backlinko, We Analyzed 11.8 Million Google Search Results (April 2025)
- Backlinko and BuzzSumo, We Analyzed 912 Million Blog Posts (2019)
- Orbit Media Studios, Blogging Statistics and Trends: The 2026 Survey of 1,042 Bloggers (2026)
- Ahrefs, Short vs. Long Content in AI Overviews (December 2025)
- Google Search Central, Creating Helpful, Reliable, People-First Content (2026)
- Nielsen Norman Group, How Little Do Users Read? (May 2008)
- Chartbeat via Time, What You Think You Know About the Web Is Wrong (March 2014)
- Contentsquare, Digital Experience Benchmark 2026 (2026)
- Wistia, What Is the Optimal Video Length? (State of Video 2026 data) (2026)
- Vidyard, Business Video Benchmarks (2025)
- DataReportal, Digital 2026 Global Overview Report (October 2025)
- Reuters Institute for the Study of Journalism, Digital News Report 2026 Executive Summary (2026)
- Pew Research Center, Social Media and News Fact Sheet (2025)
- Socialinsider, Social Media Benchmarks (2026)
- Buffer, Social Media Benchmarks (January 2026)
- Semrush, AI Overviews Study (2025)
Methodology. Numbers were collected in September 2026 and checked against the original publisher's page, not against statistics roundups. Vendor claims without a stated sample or method were excluded. Word count studies are reported as correlations, because none of the publishers claim causation and Backlinko says so explicitly. Medians and averages are labelled separately wherever a source publishes both. Three sources older than 24 months are marked DATED and kept because nothing larger has replaced them: Backlinko and BuzzSumo on links (2019), Chartbeat on attention (2014) and Nielsen Norman Group on reading (2008). Last updated 20 September 2026.