AirOps analyzed 815,484 pages and found the citation sweet spot is 500 to 2,000 words. The Digital Bloom analyzed 30 million citations and found 10,000-word pages get 62 times more citations than shorter equivalents. Both studies are correct. They are measuring different things, and understanding the difference determines how you structure content for AI search.
What did AirOps's study of 815,484 pages find about content length?
AirOps found that pages between 500 and 2,000 words with 7 to 20 subheadings earn the most consistent ChatGPT citation rate.
The finding comes from an AirOps analysis published in April 2026, covering 815,484 retrieved pages and 16,851 ChatGPT queries, measuring citation rate — defined as the percentage of retrieved pages that end up cited in the final answer.
The study also identifies what it calls the "ultimate guide paradox": long-form comprehensive pages with the highest word counts, the most headings, and the highest domain authority in the dataset were among the least reliable performers by citation rate.
High-authority long-form pages are retrieved often but cited at a lower rate once retrieved.
ChatGPT retrieves roughly 6 to 7 pages for every one it cites, discarding 85% of retrieved content before writing the answer.
A 10,000-word page covering everything is harder for a model to extract a precise, targeted answer from. A narrowly scoped 1,200-word article that resolves one question cleanly is easier to cite for that specific question.
What did The Digital Bloom's study of 30 million citations find about content length?
The Digital Bloom analyzed 30 million AI citations in December 2024 to assess content format performance. The study held topic constant, comparing a 10,000-word piece with a Flesch Reading Ease score of approximately 55 against a shorter equivalent covering the same subject, so length and reading level were the only variables that changed.
The long-form piece earned 187 total citations against 3 for the shorter equivalent — 62 times more citations in raw count.
The 187-versus-3 figure is a citation volume count, not a per-query rate — it reflects how many times a page was cited across everything it was retrieved for, not how often a single retrieval converted into a citation.
Why did AirOps and The Digital Bloom reach opposite conclusions on content length?
AirOps and The Digital Bloom disagree because they are measuring different things: citation rate per query, versus total citation volume accumulated over time.
AirOps measures citation rate per query — for any single question, what share of retrieved pages get cited. That metric rewards focused, extractable content, which is why its sweet spot lands at 500 to 2,000 words.
Total citation volume is what The Digital Bloom measures instead, accumulated across many queries over time. A comprehensive 10,000-word guide can answer not just one query but 20 to 30 distinct sub-questions, and each time ChatGPT, Perplexity, or Google AI Overviews responds to any one of those sub-topics, the long-form page is a candidate.
| Metric | Favors shorter content (500-2,000 words) | Favors longer content (10,000+ words) |
|---|---|---|
| Citation rate per query | Higher | Lower |
| Total citations over time | Lower | Higher |
| Number of queries covered | Fewer | Many |
| Extractability per section | Higher | Variable |
| Domain authority signal | Neutral | Stronger over time |
Source: AirOps (n=815,484), The Digital Bloom (n=30M+ citations)
A tightly scoped page built around a single question — for example, "does author attribution affect ChatGPT citations" — will win the per-query citation race outright. A comprehensive guide on AI search optimisation will accumulate more total citations over six months by being retrievable across dozens of related sub-queries.
Both are valid. The choice depends on whether you are chasing one query's citation rate or building topical authority across a subject.
The readability constraint both studies agree on
AirOps analyzed 353,799 pages for readability and found the citation peak at Flesch-Kincaid grade level 16 to 17 — a 35.9% citation rate. This is college-level writing: complex enough to demonstrate expertise, structured enough to stay extractable.
Pages scoring 50 or above on the Flesch Reading Ease scale also appeared more frequently in ChatGPT citations. That range spans "fairly difficult" to "standard" on the Flesch scale — the register used in industry publications and research summaries, not simplified web copy.
The finding cuts against the common advice to write for a 6th-grade reading level for content discovery. AI systems do not reward oversimplification — they reward precision: specific terminology used correctly, sources cited explicitly, claims stated with appropriate qualification.
The format that works across both dimensions
The format that reconciles both studies is the category-level page — one URL covering multiple intents instead of a single narrow angle.
Growth Memo's analysis of 21,482 ChatGPT citations found 58% of cited URLs are cited only once, while the top 4.8% — cited 10 or more times each — were all category-level pages covering what the topic is, who uses it, how to choose, and what it costs, all in one place.
Long-form, multi-intent coverage drives the high total citation volume; structuring it as modular H2 sections, each independently extractable, is what adds the per-query citation rate on top. That combination outperforms both pure long-form and pure short-form pages.
Microsoft documented this directly: strong descriptive headings are "signals that help AI know where a complete idea starts and ends."
Each H2 section should be independently understandable without context from surrounding sections — write each section as if it is the only thing an AI system will read, because for any given query, it may be.
How do you decide between short-form and long-form content for AI citations?
Decide based on your goal — write a focused piece to target a single query, or a long-form modular guide to compound citations across a topic.
For a focused piece, aim for 500-2,000 words with 7-20 subheadings, each section answering one clear sub-question. Prioritise extractability over comprehensiveness.
For building citation volume across many related queries, write long-form with each H2 section built as an independently extractable unit, and aim for at least 50 on Flesch Reading Ease.
Include a visible publication date and at least one specific number in the first paragraph. Growth Memo found that DATE and NUMBER are the two strongest positive entity signals in a page's first 1,000 characters for AI citation selection.
The mistake to avoid is writing long content without structural clarity.
Long and unstructured content produces worse per-query citation rates than short, focused content — length only pays off when the headings create genuine extraction points for AI systems.