Short articles get cited without meeting a universal word count
Short articles receive AI search citations. In an Ahrefs study of Google AI Overviews, 53.4% of cited pages contained fewer than 1,000 words. Citation research gives you little reason to expand every short article simply to make it longer.
A separate study found higher average citation counts for longer articles, but it did not test whether adding words to the same article increased citations. Base your editing decision on whether the article answers the reader's question. Keep a short article concise when its explanation is complete; add pricing, specifications, or instructions when that information is missing.
Pages under 1,000 words made up 53.4% of AI Overview sources
On December 3, 2025, Ahrefs published an analysis of 174,048 pages cited in AI Overviews. It counted words on pages whose content it could extract, removing HTML, navigation, and other boilerplate.
The cited pages fell into these length groups.
| Page length | Share of analyzed cited pages |
|---|---|
| Under 1,000 words | 53.4% |
| 1,000–2,000 words | 30.6% |
| Over 2,000 words | 16.0% |
The denominator is the 174,048 cited pages analyzed. Of those pages, 53.4% contained fewer than 1,000 words. (Ahrefs' AI Overview content length study)
That percentage describes the sources selected. If short articles are more common to begin with, they could also make up a larger share of citations. The study examined cited pages only, leaving open which length group had the higher likelihood of being cited.
The sample also included pages other than blog posts. Median word counts were 317 for core pages such as homepages, About pages, and service pages, versus 1,166 for articles. These different page types serve different purposes, so the overall 53.4% figure gives you no basis for shortening every blog post. (Ahrefs' breakdown by page type)
The published text does not specify the location, query language, or data collection period.
Longer articles averaged about 1.6 times as many ChatGPT citations
On November 24, 2025, SE Ranking published research on ChatGPT citations and website characteristics. It analyzed 129,000 domains and 216,524 pages across 20 niches, using data from 100,000 prompts.
It reported the following average citation counts for shorter and longer articles.
| Article length | Reported average citations |
|---|---|
| Under 800 words | 3.2 |
| Over 2,900 words | 5.1 |
Longer articles averaged about 1.6 times as many citations as shorter articles. Articles under 800 words still averaged 3.2 citations, providing further evidence that short content can be cited. (SE Ranking's ChatGPT citation study)
Longer articles may also differ in depth, supporting evidence, or the websites publishing them. The study compared existing articles rather than measuring the same articles before and after expansion. It therefore leaves the effect of adding words untested.
The published text does not specify the location, prompt language, collection period, or number of pages in each word-count group.
Expand articles that leave the reader's question unanswered
When reviewing a short article, compare its content with the question it is meant to answer. A product comparison should explain relevant feature and cost differences; an implementation guide should explain setup and migration. Add information needed to answer the question. Keep the length when those explanations are already complete.
Google's helpful content guidance asks whether readers learn enough to achieve their goal. Use that test when reviewing an article, rather than writing to a presumed search-engine word count. (Google's helpful content guidance)
| Missing explanation | Information to add |
|---|---|
| A feature is described without explaining which plans include it | Eligible plans and any additional fees |
| Implementation is discussed without explaining migration | Supported data, migration steps, and unsupported formats |
| Results are reported without a measurement method | User count, measurement period, tasks compared, and calculation method |
| Products are listed without reasons to choose them | Differences by use case, supported by specifications or testing |
Our guide to query fan-out and choosing information to add to an article also explains how to identify relevant questions to answer.
Remove repetition and off-topic explanations from longer articles
Cut repeated conclusions and general background that does little to answer the question. A lengthy history of a product category, for example, adds little to a comparison of current prices.
Keep supporting evidence and feature restrictions. Removing plan eligibility or usage limits while retaining a claim that an integration is available can imply unrestricted access on every plan. Preserve the details readers need to determine whether a product fits their requirements.
For longer articles, give the answer near the beginning, then explain the reasons and details under descriptive headings. Help readers reach the information they need while keeping essential explanations. Our guide to answer-first content explains this approach in more detail.
Check cited URLs using the same questions after editing
Before and after editing, repeat the same questions in the same AI service. Keep the language and location consistent, and record which URLs from your site and competitors appear as sources.
Record the editing date and the pricing details, specifications, or test results you added. If articles gaining citations share similar additions, investigate whether those explanations are missing elsewhere on your site. AI updates can also change answers, so compare results over time. Our guide to measuring AI search visibility explains the measurement process.
Find gaps in article explanations with Please Cite Me!
Short articles receive AI citations, so length alone gives you little reason to expand them. Check whether the article answers its intended question, then fill gaps in explanations or evidence. In longer articles, retain necessary information and remove repetition or unrelated material.
Please Cite Me! is a reporting tool for SEO agencies that checks whether a page is publicly accessible and whether its content is sufficiently developed, then produces an audit report.
When revising an article, try Please Cite Me! to check page accessibility and identify gaps in its explanations!
About the author
Yuma Okamoto
Developer of Please Cite Me! and CEO of Abeam Tech Inc.
University of Tokyo graduate focused on web product development, SEO, and information architecture for AI search, with a particular interest in how web content is understood and cited by search engines and AI.
X / @okamoto_yuma_Frequently asked questions
What is the ideal content length for AI citations?
These studies establish no universal ideal word count. Short articles also receive citations. Judge length by whether an article answers its intended question and includes the evidence and qualifications readers need.
Should I rewrite short articles to make them longer?
Expand them when information needed to answer the question is missing, such as pricing, specifications, or implementation steps. If the explanation is complete, keep it current and continue monitoring citations.
Should I replace a long article with a summary?
Keep the answer together with its reasoning and verifiable evidence. Removing eligible plans, usage limits, or a study's measurement period can change the meaning. Cut repetition and unrelated material, then organize necessary details under descriptive headings.
Sources
Please Cite Me!
Audit your page for AI search
Please Cite Me! reviews a public page's technical access and editorial clarity, then turns the evidence into a prioritized report.
Audit a page