1. The Concept, Plainly
The format of your content — whether it is a review, a how-to guide, a product page, or a thought leadership piece — shapes whether AI systems cite it. A 2025 practitioner study found that citation share across content types is sharply unequal: some formats attract the majority of AI citations, while others attract almost none. This is separate from how well a piece is written or how much evidence it contains.
2. Why This Matters Right Now
Posts L1 through L4 in this series have covered structural properties — evidence density, source attribution, readability, recency signals. Those concepts operate inside a piece of content: they improve the properties of what you have already decided to produce.
L5 introduces a prior question: is the type of content you are producing competitive for AI citations in the context you are targeting?
The answer varies sharply by format. If your content type draws a small share of citations in your target context, structural improvements alone cannot close that gap — you are building quality into a format that is already at a disadvantage.
3. The Mechanism
A 2025 practitioner study by Omniscient Digital analysed 23,387 unique sources drawn from 240 branded-query prompts — each containing a specific brand name — across five AI platforms. The dataset is large and the methodology is stated. It is not peer-reviewed. Treat the figures as directional signals, not certified benchmarks.
Across those 23,387 citations, the distribution by content type was not even close:
- Reviews, listicles, forums, and case studies: approximately 57% of citations
- Directory sites: approximately 17%
- Product pages: approximately 12%
- Thought leadership: approximately 5.4%
- Video and news/press releases: the least-cited formats
A companion study from the same research programme extended the analysis to 43,000+ citations and added an intent-stage dimension:
- At the Problem Unaware stage — queries from users who have not yet identified a solution — educational and thought leadership content accounted for approximately 86% of citations
- At the Solution Aware stage — queries from users comparing options — reviews, comparisons, and community content accounted for approximately 51% of citations
Thought leadership is not disadvantaged in all contexts — at the awareness stage, the picture reverses. The same content type can dominate one query context and be marginalised in another.
What the data cannot tell you — stated plainly.
Every prompt in the Omniscient dataset contained a brand name. If you produce content for query contexts where no brand name appears, the percentages do not transfer directly to your situation.
The mechanism is also untested. Whether the distribution reflects AI weighting of content types, a composition effect in the source pool, or something specific to branded-query contexts is an open question. The data describes distribution. It does not establish cause.
The Layer 1 / Layer 2 framework.
The research from earlier posts — Aggarwal et al. and Zhang et al. — operates at Layer 2: the structural properties inside a piece of content. Evidence density, source attribution, heading structure, readability. That research is peer-reviewed.
The Omniscient finding adds a prior layer. Call it Layer 1: given the query context you are targeting, is your content format competing for a meaningful share of citations at all?
Layer 1 is the format question. Layer 2 is the quality question. Structural optimisation is solving a real problem — but only inside the constraint the format has already set. Layer 2 work matters. It matters more when Layer 1 is answered first.
4. Try It Now
No dedicated content-type tool exists on Psytable for this step — the exercise below is prompt-only. The prompt asks an AI to classify your content by type and intent stage, then compare that against the citation distribution. The output is a format question, not a score.
Paste this prompt:
"Here is a description of the content I produce [or: here is a short excerpt]. Please do three things. First: identify the content type — thought leadership, review, comparison, case study, forum post, how-to guide, directory entry, product page, news release, or another type if none of these fit. Second: assess which query intent stage this type best serves — Problem Unaware (the reader has not yet identified a need) or Solution Aware (the reader is comparing options) — and explain briefly why. Third: a 2025 practitioner dataset of 23,387 branded-query citations found thought leadership at approximately 5.4% citation share, reviews and listicles at approximately 57%, and educational content at approximately 86% of citations for Problem Unaware queries. Given those figures, does this content type appear competitive for the intent stage you identified, or is there a mismatch? Note: the dataset covers branded queries only and the percentages may not transfer to informational content.
[Describe your content type in one or two sentences, or paste 100–200 words of an example]"
What to look for in the output: The AI will classify your content type and identify the intent stage it serves — this is your Layer 1 reading. A Layer 1 match is alignment: educational content targeting Problem Unaware readers sits in the 86% citation share zone. A Layer 1 mismatch is when your format and context point in opposite directions — thought leadership aimed at vendor-evaluation readers, for example, where that format averages 5.4% citation share — and that mismatch is the question the data raises, not an instruction to change format.
5. The One Thing to Remember
The format of your content shapes whether it competes for AI citations in your target context — and structural optimisation cannot override a Layer 1 mismatch between your format and that context.
6. Go Deeper
The Field Notes post on this study covers the full distribution data, the branded-query scope limitation, and how the Layer 1 / Layer 2 framework applies to sequencing your optimisation work: In Branded-Query Data, Content Type Correlates with Citation Share.
Measure Layer 2 once Layer 1 is answered.
The Evidence Density Score measures statistics presence, source attribution, readability, and structural richness — the Layer 2 properties associated with citation probability in the Aggarwal and Zhang research.