Brand standards, photography budgets, and editorial taste have generally guided visual strategy in PR and content marketing. Data on which images actually drive results has been harder to come by.
The Fullintel research on AI citation behavior, conducted with Dr. Tyler Page from the University of Connecticut, offers a specific, statistically significant answer for earned media: articles featuring people rather than products are cited more consistently by AI search engines.
In 63% of news articles in the cited sample, extracted from 6,183 URLs from ChatGPT, Gemini, and Google AI Overview outputs, featured people as the primary visual element rather than products or branded graphics. Human visuals without brand prominence predicted AI Responses (beta = .077, p < .05) and Citation Consistency (beta = .094, p < .01) in the regression models. The association was statistically significant and replicated across platforms.
For PR teams with any influence over editorial photography or content visuals, that finding is actionable.
What the Data Actually Shows
The research coded primary imagery in articles across several categories: people-focused, product-focused, branded graphics, and abstract or stock imagery. The 63% figure reflects the proportion of articles in the cited set that used people-focused primary imagery. That is the entry-level observation.
The regression analysis goes further. It tests whether human imagery, after controlling for other variables, is independently associated with more frequent AI citation. It is. Human visuals without prominent brand elements predicted both how often an article appeared in AI responses and how consistently it was reused across different prompts and personas.
Critically, product shots and branded graphics did not show the same positive association. This is not a finding about which images look better or perform better with human audiences. It is a finding about how AI retrieval systems appear to weigh articles when selecting and reusing them.
This finding is exciting because I have done research that shows that humans engage with human visuals (p. 152), but it hasn’t been tested with machines. This is a preliminary study to confirm that AI search engines also prioritize content with human visuals.
Why This Pattern Might Exist
The research does not establish a causal mechanism, but the correlation is consistent enough to suggest a plausible explanation. Human-centered imagery is associated with editorial content that prioritizes people, stories, and context over promotional framing. AI retrieval systems may encode that association as a trust signal.
An article illustrated with an authentic photograph of a healthcare professional discussing treatment options reads differently from one illustrated with a product package shot or a branded logo on a brick building. The first signals editorial independence and contextual reporting. The second signals proximity to promotional content.
Retrieval systems trained on large corpora of web content likely encode what authoritative, trustworthy content looks like at the structural and metadata levels. Human imagery, particularly imagery without prominent brand elements, may be one such structural signal.
What PR Teams Can and Cannot Control
PR professionals rarely control the photography choices of the journalists and editors who cover their organizations. That limitation is real. But influence over visual strategy is not zero.
Contributed articles and op-eds with the brand’s direct editorial input are one opportunity. If you are placing a bylined piece by a company executive or thought leader, the choice to use an authentic headshot or a real-world contextual photo rather than a branded product image is entirely within your control.
Press materials represent another lever. The images you provide in press kits and embargoed materials influence editorial choices. If your default press photography is product-centric, consider whether adding people-focused alternatives, executives in working environments, practitioners in context, or customers in authentic settings changes the coverage you receive.
Owned content on high-authority properties is a third area. Blog posts, white papers, and resource pages on your own domain that qualify as high-authority are also part of the AI citation landscape. Visual choices on those properties are entirely under your control.
The Authenticity Variable
While human visuals without brand prominence was the most influential factor, any images with humans whether branded or not are preferred over visuals without human elements. The recommendation would be to always integrate human visuals, and if possible include brand in an authentic way. Humans engage with humans, and humans are more likely to remember brands if they are in the visuals. This approach maximzes both AI search engine citation and human recall that can shape perception and behavior.
Authentic, contextual photography, real people in real situations without prominent brand staging, is the category associated with AI citation. Stock photography of anonymous individuals in generic settings is likely to behave differently from authentic imagery of named individuals in identifiable professional contexts.
This reinforces a broader direction in communications: authenticity in visual storytelling serves multiple goals simultaneously. It builds trust with human audiences, meets editorial standards, and, according to this research, appears to support AI citation behavior.
Connecting Visual Strategy to Coverage Measurement
Most PR measurement frameworks do not track visual characteristics of earned media at scale. Knowing which articles feature people versus products requires a level of content coding that manual review rarely supports.
The immediate application is simpler: apply this lens to your own content and press materials, and track whether coverage from pitches with people-focused photography differs from coverage from product or brand-led materials. The research gives you a hypothesis worth testing in your own program.
Fullintel’s media intelligence team supports communications leaders who want to understand their earned media portfolio at the content-level, not just volume and sentiment. If you are building toward AI citation as a measurement outcome, visual strategy is one of the inputs worth tracking. Learn more about our strategic media analysis capabilities.
