A survey of 500 marketers and business owners published by NP Digital in May 2026 has quantified for the first time which content formats earn citations in AI-powered search results - and the gap between top and bottom is enormous. Original research scores 82%, video content scores 2%, and almost every widely produced content type sits closer to the bottom than the top.

The findings, shared today via email by Neil Patel, co-founder of NP Digital, carry direct implications for marketing teams that have continued to invest in content formats that AI systems can produce without attribution. The survey, conducted online with a base of 500 marketers and business owners, was published in May 2026 and covers 11 distinct content categories. It presents one of the most direct datasets to date on how AI search engines select what to cite - and what to ignore.

Why this ranking matters now

Google's search interface has shifted. According to NP Digital, the platform is no longer a traditional search engine with 10 blue links but rather an AI-first system where the AI response appears first and traditional results appear below it. That structural change means the question of whether content earns a citation inside an AI response has become a more consequential measure than whether a page ranks in position one.

The shift has been documented in detail. Research tracked by PPC Land found that when AI Overviews appear, click-through rates at position one fall from 27% to 11%. In Germany alone, across 100 million keywords, that compression amounts to 265 million lost organic clicks per month. A February 2026 Ahrefs study covering 300,000 keywords found that AI Overviews correlate with a 58% reduction in click-through rates for top-ranking pages - nearly double the 34.5% figure the same company documented in April 2025. Into that environment, the NP Digital survey lands with a specific and usable question answered: if citation is the new visibility, which content earns it?

The ranking in full

The survey ranks 11 content types from highest to lowest AI citation performance:

Original research scores 82%. Comparison content scores 76%. Rankings and best lists score 57%. FAQs score 41%. How-to content scores 39%. Community and forum participation scores 28%. Generic blog posts score 25%. Definition and explanation pages score 22%. Opinion and thought leadership scores 16%. Product pages score 14%. Video content scores 2%.

The distance from first to last - 80 percentage points - is larger than it might first appear. It is not a gentle slope. It is a cliff between the top two formats and everything else, followed by a much slower decline through the middle, and then a second cliff at the bottom where video sits almost entirely alone.

The structural logic behind the numbers

According to NP Digital, the two top-performing formats share a defining characteristic: they contain information that AI engines cannot generate without a source. Original research at 82% and comparison content at 76% both require data, methodology, and conclusions that exist nowhere in an AI system's training data unless someone has published them. Every other content type in the ranking, by contrast, contains information AI can produce from its training data, which is why those formats get cited less - citation is a necessity for original research, but a choice for everything else.

"Original research and comparison content are what AI engines want to cite because they contain information AI cannot generate itself," said Neil Patel. "If all your content explains things AI already knows, you are producing content AI can summarize without mentioning you. Be the source, not the summary."

The logic extends further into the comparison content category. According to NP Digital, a comparison of five CRM platforms based on proprietary testing criteria is content AI cannot replicate without referencing the source. Generic comparisons without proprietary data or unique evaluation criteria, however, are expected to perform closer to the generic blog post category at 25%. The label "comparison content" does not automatically confer advantage - the proprietary methodology is the ingredient that earns the citation.

The drop from comparison content at 76% to rankings and best lists at 57% reflects a meaningful decrease in content specificity. Rankings are more replicable than comparisons. FAQs, at 41%, contain question-and-answer formats that AI engines can construct independently. These formats still contribute to AI citation rates, particularly when they incorporate specific examples or proprietary methodology, but they do not generate citations as reliably.

Generic content and the training data problem

Generic blog posts at 25% and product pages at 14% rank near the bottom because they represent the formats AI engines can reproduce most completely without needing to reference the original. A generic explanation of what content marketing is, or a product description of a standard software tool, is information an AI system already has in its training data. Publishing more of it does not generate citations. It generates content that AI can summarize without attribution.

This finding has a specific operational consequence. According to NP Digital, most content teams allocate a disproportionate share of their production budget to generic blog posts and definition pages because they are faster and cheaper to produce. The volume benefits of that approach no longer translate to citation benefits - and in an environment where citation determines AI visibility, the cost calculation has inverted.

PPC Land has tracked how the gap between organic ranking and AI citation has become a practical measurement problem for marketing teams. Research from Ahrefs in June 2025 added a conversion dimension: AI search traffic converts at 23 times the rate of traditional organic traffic, despite representing only 0.5% of total visits. That ratio makes citation more commercially meaningful than the raw traffic numbers suggest.

Video at 2%: a technical limitation, not a permanent verdict

Video content's last-place ranking at 2% is explained differently from the rest of the list. According to NP Digital, the figure reflects a current technical limitation rather than a permanent ceiling. AI engines process text more reliably than video, which is why video content is rarely cited even when it contains genuinely original information. The analysis notes that as multimodal AI capabilities mature and video transcripts become more systematically integrated into AI training data, this figure will likely improve.

That is a meaningful qualifier. The 2% score applies to video as AI search engines process it today - not to the informational quality or originality of video content as a format. A proprietary interview or original documentary that exists only in video form may contain citation-worthy information, but AI systems currently lack a reliable mechanism to extract and attribute it.

Community and forum participation: the underweighted channel

Community and forum participation scores 28% - ranking above generic blog posts and definition pages, yet, according to NP Digital, most content teams do not include it on their editorial calendar at all. The category captures the peer-discussion signals that AI engines interpret as third-party validation of expertise. Expert answers in relevant industry forums, Reddit communities, LinkedIn discussions, and Quora topics represent contributions that appear in AI training data and build the kind of distributed presence that editorial content produced on owned domains cannot replicate on its own.

PPC Land reported in June 2025 how Aleyda Solis released a comprehensive AI search content optimization checklist covering eight distinct optimization areas, including authoritativeness signals such as expert bylines, structured data implementation, external citations, and mentions on reputable websites. The NP Digital ranking reinforces that framing: forum participation scores higher than generic blog posts not because of where it is published but because of the structural signals it sends to AI systems about expertise and peer validation.

Opinion and thought leadership at 16%

Opinion and thought leadership content scores 16%, placing ninth in the ranking. That figure warrants specific attention because the category sounds like it should perform better - original perspectives, named authors, explicit points of view. The data suggests otherwise. The issue is that opinions, unless grounded in original data or proprietary analysis, remain replicable. An AI system can construct a plausible opinion on most marketing topics from its training data. Thought leadership that scores well would need to be attached to original research or proprietary comparison data to cross into the top two categories.

A May 2026 analysis by Cyrus Shepard covering 54 experiments, patents, and case studies, covered by PPC Land, placed URL accessibility and search rank as the top two AI citation factors, scoring 9.5 and 9.4 respectively on an evidence-weighted scale. That finding aligns with the NP Digital survey's implicit conclusion: traditional SEO fundamentals still matter, but the content format has become an independent variable in whether a citation occurs at all.

Industry context: most marketers are not adapting

The NP Digital survey arrives at a moment when documented evidence of marketer unpreparedness for AI search is accumulating. According to Adobe, 98% of marketers lack a clear, documented roadmap and full confidence in their AI optimization approach. Of those surveyed by Adobe in a separate study covering more than 500 marketers, 74% either have no measurable strategy for AI search and LLM discovery at all, or are unaware of one within their organization. Adobe published those findings on April 9, 2026.

PPC Land covered how Lily Ray, in May 2026, identified specific risks of aggressive GEO tactics - including the possibility that optimizing specifically for AI citation rather than quality could trigger search penalties that compound across both traditional and AI search surfaces simultaneously. The NP Digital survey's recommendation - focus on content AI cannot generate itself - sidesteps that risk by aligning the citation strategy with quality rather than with tactical manipulation.

Google published its own guidance on May 15, 2026, taking the position that optimizing for generative AI search is optimizing for the search experience, and is therefore still SEO. That framing aligns with the NP Digital data: original research and comparison content earn citations not because they trick AI systems but because they contain information those systems genuinely need to attribute.

Implications for content investment decisions

The NP Digital survey is explicit that it functions as a content investment directive. If a content production budget is weighted toward formats AI can replicate, the citation rate will reflect that. The data does not recommend eliminating all lower-performing formats - FAQs at 41% and how-to content at 39% still contribute meaningfully, particularly when they incorporate specific examples or proprietary data. What it suggests is that the production mix at most organizations is inverted relative to what the citation environment rewards.

PPC Land has documented how the SEO industry has grappled with new acronyms - AEO, GEO, AIO - as practitioners try to name the practices involved in improving AI visibility. The NP Digital data offers a simpler frame: the content types that AI systems must cite are the content types no AI system can produce on its own. The practical implication is that investment in original research and proprietary comparison content is the most direct path to AI citation - not because of how the content is structured or tagged, but because of what it contains.

Timeline

Summary

Who: NP Digital, the digital marketing agency co-founded by Neil Patel, and its survey base of 500 marketers and business owners.

What: A survey ranking 11 content types by their performance in AI search, defined as the rate at which each format earns citations in AI-generated search results. Original research ranks first at 82%. Comparison content ranks second at 76%. Rankings and best lists rank third at 57%. Video content ranks last at 2%.

When: The survey was conducted and published in May 2026. Neil Patel distributed the findings today, June 17, 2026, via email to marketing professionals.

Where: The survey was conducted online with 500 marketers and business owners. The findings were published on the NP Digital website under the content strategy section and distributed via email marketing.

Why: The ranking matters because AI search engines have become the primary interface through which users receive answers on Google and other platforms, making AI citation a more consequential measure of content visibility than traditional organic rankings. Content that AI systems can produce without attribution is increasingly invisible; content that AI systems must cite is increasingly valuable. The survey quantifies that distinction across 11 formats, giving marketing teams a direct framework for adjusting content production priorities.