ChatGPT Shopping uses different operational mechanisms and ranking criteria for product recommendations in the United States compared to Europe, according to new research from Peec AI.
The distinction means merchants must adopt tailored strategies to ensure their products appear in AI-generated shopping results, with implications for e-commerce visibility across continents.
In the United States, ChatGPT Shopping utilizes OpenAI’s proprietary Labrador index, which is directly populated by product feeds, Peec AI reported. This system mandates a verified seller status for any store to be included in search results, effectively excluding unverified merchants.
Conversely, European ChatGPT Shopping results are retrieved from Google Shopping via a third-party application programming interface (API), according to the analysis. This approach means European product listings are subject to Google’s ranking factors rather than OpenAI’s internal system.
Metehan Yesilyurt, an AI researcher at Peec AI, stated that the U.S. system relies on a complex relevance scoring mechanism. This scoring incorporates embeddings, BM25, and GTIN (Global Trade Item Number) matching to determine product placement.
In contrast, European results are rotated without such detailed relevance scoring, said Malte Landwehr, CEO of Peec AI. He added that this rotational display means a product’s appearance is less about specific ranking factors and more about inclusion in Google Shopping.
A critical factor for U.S. merchants is how ChatGPT processes product descriptions. The AI truncates descriptions to approximately 50 tokens, or about 40 English words, for scoring purposes. This makes the early placement of key information in product descriptions essential for ranking, Peec AI advised.
For optimal visibility in the U.S., Peec AI recommends that companies like Nike and Dick’s Sporting Goods send product feeds directly to OpenAI and declare each product variant separately. This ensures the Labrador index can properly categorize and score items.
European merchants, including those using platforms such as Shopify, Bright Data, Salesforce, and Stripe, should focus on optimizing their listings for Google Shopping until OpenAI’s Labrador index is deployed in the region, Peec AI suggested.
The research highlights a significant divergence in how AI-powered shopping recommendations are generated, compelling e-commerce businesses to adapt their digital marketing and product feed management strategies based on geographic target markets.
Source: Abondance
