AI Tools Recommend Brands But Cite Other Sites, Data Shows

Aug 27, 2026 04:13 AM - 1 hour ago 4

AI devices often propose brands without straight linking to their charismatic websites. Shero Commerce’s caller analysis recovered that only 2.8% of the 1,851 sources cited crossed Google AI Mode, ChatGPT, and Perplexity were brand-owned pages.

Shero, a Shopify agency specializing successful SEO and AI services, looked into citation trends, but the study doesn’t show why AI systems favour immoderate sources complete others.

Third-Party Sites Received Most Citations

Shero examined purchasing questions crossed 60 merchandise categories, gathering each sources provided by 3 different AI tools. Most references pointed to websites extracurricular of brand-owned domains. The biggest root class made up 59% of citations, including sites for illustration Good Housekeeping, Verywell Fit, and Reviewed.com. Brand-owned pages accounted for only 2.8%.

This shape besides appeared erstwhile AI devices suggested brands by name. Shero recovered 159 marque recommendations crossed each platforms, pinch the brand’s ain page cited successful 31% of those cases.. For example, some ChatGPT and Perplexity recommended brands specified arsenic Gymshark, Alo, and Beyond Yoga for a squat-proof leggings query, citing third-party sites.

Google AI Mode Rarely Named Sampled Brands

Shero recovered that successful its sample, brands were cited aliases recommended successful 9.5% of applicable Google AI Mode shop checks crossed 60 different categories. In astir one-third of these categories, nary sampled brands were present.

The study besides mentions that celebrated devices for illustration ChatGPT and Perplexity suggested well-known brands, but Shero didn’t connection a comparable proposal complaint for these platforms. So, while the 9.5% fig from Google is useful independently, it shouldn’t beryllium straight compared to different platforms.

Why This Might Be Happening

The citation study was portion of a broader reappraisal of merchandise contented crossed Shopify stores. In checking a stratified sample of merchandise descriptions, Shero recovered 20% were identical to aliases very akin to those connected different websites. These similarities mostly appeared among retailers aliases marketplaces reselling the aforesaid products, alternatively than betwixt competing Shopify brands.

The authors propose that syndication could make attribution much difficult erstwhile identical matter exists crossed aggregate sites. However, the study doesn’t show that plagiarism resulted successful third-party sites receiving much citations.

The study besides highlighted that short Shopify explanation fields whitethorn underestimate the existent magnitude of matter connected a page. Out of 173 stores Shero could cleanly measurement utilizing earthy HTML, 27 had less than 50 words of product-specific content.

Shero collected 8,573 merchandise descriptions from 883 stores pinch unrecorded merchandise data.

Why This Matters

For ecommerce brands, being recommended and being cited are 2 abstracted outcomes. One brand whitethorn beryllium highlighted successful an AI buying answer, while different tract mightiness person the root link. Shero’s information offers immoderate penetration into the difference, but doesn’t corroborate why AI devices prime those peculiar sources.

Looking Ahead

The main point to see is whether changing the merchandise contented affects citation behavior. Shero didn’t trial if rewriting syndicated descriptions, adding original copy, aliases adjusting page building makes it much apt for a brand’s page to beryllium cited. To really understand this, a circumstantial trial focusing connected those changes would beryllium necessary. For now, the study conscionable shows wherever citations occurred, without explaining why those sources were selected.

Category News AI Search
Add SEJ arsenic a preferred root connected Google
More