Jellyfish Research Exposes Fractured AI Shopping Shelf Brands Cannot See
The Development
Brandtech Group's Jellyfish has published research and launched a new capability within its Share of Model platform — Shopping Optimization — that maps how AI shopping assistants evaluate, rank, and recommend individual SKUs across ChatGPT, Google AI Mode, and Amazon's Alexa for Shopping. The research, spanning eight product categories across the US, UK, Australia, and Singapore, found that ChatGPT accounts for approximately 80% of AI product recommendations versus Google AI Mode's 20% — a ratio that reaches 97-to-3 in home appliances. A single query surfaced as many as 290 competing brands in one category. In UK men's suits, price ranges within a single AI recommendation spanned £27 to £3,295. Asked about toys, Amazon drew on 177 brands but exclusively from its own store; ChatGPT sourced from 24 retailers and Google AI Mode from 37.
Our Take
The research makes explicit what many brand leaders have been avoiding: the category hierarchies and price-tier defences brands spent decades constructing are not legible to AI shopping systems. A £27 suit and a £3,295 suit occupy the same recommendation response. The dominant brand in a category may hold just 7% of the AI shelf. More structurally threatening is the assistant fragmentation — winning on ChatGPT tells a brand almost nothing about its position on Google AI Mode or Alexa. Brands that treat AI visibility as a single optimisation problem are operating on a false model. The measurement challenge is genuinely new: Share of Model Shopping Optimization is attempting to instrument a moving target across competing AI systems with different data sources and retailer relationships.
What Changed
For the first time, marketers can analyse AI shopping behaviour at individual SKU level — not just brand mentions — identifying which product attributes, content signals, and retailer relationships drive recommendations across specific AI assistants, and tracking changes in visibility over time as optimisation efforts are applied.
Marketing Impact
Ecommerce and brand teams face an immediate measurement gap. Conventional digital shelf analytics and first-generation AI visibility tools do not capture SKU-level recommendation dynamics or retailer sourcing behaviour across AI assistants — the signals that now determine whether a product surfaces before a consumer ever reaches a retailer.
Competitive Implication
Brands with structured product data, strong retailer content programmes, and early Share of Model instrumentation gain disproportionate AI shelf presence. Brands relying on legacy search rankings and retailer shelf position as proxies for AI visibility are exposed — particularly in fragmented categories where no assistant currently has a dominant brand signal to anchor to.
Strategic Outlook
Expect competing platforms — retail media networks, marketplace analytics vendors, and GEO tooling providers — to accelerate SKU-level AI visibility products into Q1 2027. The Jellyfish launch creates category definition pressure; brands that delay instrumentation will find the optimisation gap harder to close as AI shopping assistant usage compounds.
The Exploit
Action Item
Ecommerce directors running SKUs in categories with wide AI price-range dispersion — apparel, furniture, gaming — should commission a Share of Model Shopping Optimization audit across ChatGPT and Google AI Mode before Q4 2026 peak trading to establish a baseline and identify which products are being systematically excluded from AI recommendations.