By Global Business & Technology Desk
Published: October 24, 2023
Main Facts
The rise of generative artificial intelligence and conversational search engines has fundamentally transformed how consumers discover products, compare brands, and make purchasing decisions. However, a groundbreaking new empirical study reveals a startling blind spot for digital marketers: AI-driven search platforms routinely recommend consumer brands while actively bypassing their official websites in favor of third-party domains.
According to a comprehensive analysis conducted by Shero Commerce—a specialized Shopify agency focusing on search engine optimization (SEO) and generative engine optimization (GEO)—only 2.8% of the 1,851 distinct sources cited across major AI discovery platforms were brand-owned properties. Instead, third-party publishers, review aggregators, and curated lifestyle blogs overwhelmingly capture the valuable citation links, even when the AI models explicitly mention a brand by name.
The research examined consumer purchasing queries spanning 60 distinct product categories across prominent AI search interfaces, including Google AI Mode, OpenAI’s ChatGPT, and Perplexity. The findings underscore an emerging crisis in digital visibility: being recommended by an artificial intelligence model does not equate to receiving direct web traffic. As search behavior shifts away from traditional blue links toward conversational, AI-synthesized answers, e-commerce brands face a complex new frontier in digital marketing where visibility and attribution are heavily decoupled.
Chronology
To understand how artificial intelligence models handle product recommendations and source attribution, Shero Commerce structured a multi-phased observational study executed over several months.
Phase 1: Category Selection and Query Formulation
The investigation began by establishing a representative matrix of 60 distinct product categories commonly found in modern e-commerce ecosystems, with a heavy emphasis on Shopify-hosted storefronts. Researchers formulated typical consumer purchasing questions—ranging from apparel and beauty to home goods and consumer tech—designed to trigger product recommendations within AI interfaces.
Phase 2: Data Harvesting Across AI Platforms
Between the initial formulation and final data freeze, researchers systematically fed these queries into three major AI-driven environments: Google AI Mode, ChatGPT, and Perplexity. Every source URL cited by the models, alongside every brand explicitly recommended within the conversational text output, was logged, categorized, and quantified. In total, the study captured 1,851 individual citation links across the target platforms.
Phase 3: Shopify Store and Product Content Audit
Running parallel to the AI citation tracking, Shero conducted a deep technical content audit of live Shopify stores to evaluate the underlying structural health of product pages. Researchers collected 8,573 product descriptions sourced from 883 active stores. Utilizing raw HTML parsing on a clean subset of 173 stores, the team analyzed text length, content depth, and the prevalence of duplicate phrasing across the web.
Phase 4: Data Synthesis and Correlation Analysis
In the final phase, researchers cross-referenced the AI citation patterns with the content audit data. They evaluated how frequently brands were named versus cited, investigated the prevalence of content syndication, and assessed the correlation between on-page copy length and AI citation success. The finalized dataset was published to provide the e-commerce community with an empirical baseline of how AI search engines interact with brand ecosystems.
Supporting Data
The empirical metrics compiled in the Shero Commerce report paint a sobering picture for direct-to-consumer (D2C) brands and online retailers striving for organic visibility in the age of generative search.
The Domination of Third-Party Domains
Out of the 1,851 citations analyzed across Google AI Mode, ChatGPT, and Perplexity, brand-owned domains accounted for a minuscule 2.8% of total references. The lion’s share of citations—constituting a massive 59% of all references—pointed directly to third-party lifestyle sites, publisher reviews, and consumer advocacy platforms. Dominant reference domains included well-known authorities such as Good Housekeeping, Verywell Fit, and Reviewed.com.
The Disconnect Between Naming and Linking
The study revealed a fascinating dichotomy when AI tools suggested specific brands by name. Across all evaluated platforms, researchers identified 159 explicit brand recommendations. Interestingly, when an AI model suggested a brand, the brand’s own official website was cited as the source in 31% of those cases.
For instance, when queried for recommendations regarding high-performance, "squat-proof" activewear leggings, both ChatGPT and Perplexity frequently highlighted popular brands like Gymshark, Alo, and Beyond Yoga. However, rather than pointing users to these brands’ e-commerce storefronts, the engines heavily favored linking to third-party roundups and editorial review sites that had evaluated the products.
Google AI Mode Visibility Rates
When focusing specifically on Google AI Mode, the research tracked how frequently sampled brands appeared in store-check queries across the 60 product categories. Brands were explicitly cited or recommended in just 9.5% of relevant checks. Furthermore, in approximately one-third of the tested categories, none of the sampled brands appeared at all.
Note on Platform Comparisons: While the 9.5% visibility figure offers valuable insight into Google’s ecosystem, the researchers noted that it cannot be directly compared to ChatGPT or Perplexity due to methodological variations in how those platforms surface brand mentions versus static store checks.
Content Duplication and Syndication Pressures
As part of the Shopify store audit, researchers uncovered widespread content duplication. Examining a stratified sample of product descriptions, 20% of the analyzed texts were found to be identical or heavily similar to copy hosted on other domains.
Crucially, this duplication was not occurring maliciously between competing independent Shopify brands. Instead, it was concentrated among multi-brand retailers, dropshippers, and marketplace vendors reselling identical manufacturer-supplied product lines.
Furthermore, depth of content emerged as a notable technical hurdle. Out of 173 stores that could be cleanly measured using raw HTML extraction, 27 stores featured fewer than 50 words of product-specific content on their pages. Overall, the study aggregated 8,573 descriptions from 883 live merchant sites, highlighting an industry-wide reliance on thin or syndicated text.
Official Responses and Industry Context
The release of the Shero Commerce study has ignited widespread debate across the digital marketing, SEO, and e-commerce development sectors. Industry analysts and technical SEO practitioners have weighed in on the implications of AI search behavior.
While the research data robustly maps where citations occur, the authors of the report exercised rigorous scientific restraint. The analysis does not definitively prove why AI systems systematically favor third-party domains over official brand properties, nor does it confirm whether content syndication directly causes a reduction in brand-direct citations.
"For e-commerce brands, being recommended and being cited are two entirely separate outcomes," note the report’s authors. "One brand may be highlighted in an AI buying answer, while another independent site receives the underlying source link."
Independent search engine optimization experts have noted that generative models—which rely on Retrieval-Augmented Generation (RAG)—are inherently trained to value perceived editorial neutrality, comprehensive comparison, and consensus. Because third-party review sites synthesize multiple brands into a single, cohesive article, AI algorithms frequently treat them as more authoritative information hubs than individual product pages designed explicitly for conversion.
Furthermore, digital commerce consultants point out that Shopify merchants have historically prioritized conversion rate optimization (CRO) over exhaustive informational content. Short product descriptions optimized for quick scanning and frictionless checkout may inadvertently starve AI crawlers of the rich, contextual text required to justify a direct citation.
Implications for E-Commerce Brands
The findings from Shero Commerce’s analysis carry profound strategic implications for direct-to-consumer brands, digital marketing agencies, and search engine optimization professionals navigating the transition to Generative Engine Optimization (GEO).
1. The Death of the Traditional Funnel Metric
For decades, digital marketers measured success through direct traffic, click-through rates (CTR) from search engine results pages (SERPs), and direct conversions. In an AI-first search environment, consumers may receive a complete purchasing recommendation—including brand names, pricing, and feature comparisons—without ever visiting a traditional search engine results page, let alone clicking a brand’s website. If an AI engine recommends Brand X but cites a third-party review blog, the brand gains top-of-funnel awareness but loses direct attribution and immediate first-party data capture.
2. Redefining Digital PR and Off-Page SEO
Because AI models heavily rely on trusted third-party authorities (Good Housekeeping, Reviewed.com, niche review blogs), traditional digital public relations (PR) is no longer just a brand-building exercise—it is a core technical SEO necessity. Brands must aggressively pursue placements on high-authority review sites, lifestyle publications, and comparison roundups. If an AI engine refuses to link directly to a brand’s product page, securing visibility on the third-party sites that do get cited becomes the primary path to indirect discovery.
3. The Need for Original, Context-Rich Product Content
The study’s finding that 20% of product descriptions were identical—driven largely by manufacturer syndication—signals an urgent need for content differentiation. Merchants relying on default wholesale descriptions blend into a sea of duplicate text, making it difficult for AI crawlers to establish unique value propositions. Expanding product copy beyond the standard 50-word threshold, incorporating rich schema markup, and adding proprietary editorial context, user-generated Q&A, and detailed specifications can help elevate a brand’s native pages in the eyes of LLM crawlers.
4. Future Research and Testing
Crucially, the Shero report leaves several questions unanswered. The analysis did not test whether actively rewriting syndicated descriptions, injecting original copywriting, or radically altering page architecture can successfully shift AI citation behavior from third-party sites to brand-owned domains.
To truly master Generative Engine Optimization, the e-commerce industry must now pioneer controlled experiments. Future tests must isolate variables such as structured data markup, content length, tone of voice, and digital PR saturation to determine what specific technical levers force AI models to cite the creators themselves. Until then, e-commerce brands must operate in a reality where winning the recommendation is only half the battle—getting the link remains the ultimate frontier.
