Key findings

  • Product-feed retrieval has grown roughly 15x since November 2025, reaching about 20% of shopping retrievals.
  • Nearly all (99.9%) product-feed-sourced citations land in the rank 1 position.
  • Roughly 88% of all offers still come from product detail pages (PDPs), and about 76% even for merchants with feeds.
  • ChatGPT sees completely different product metadata depending on the retrieval source.

ChatGPT retrieves product offer metadata from two, very different sources

ChatGPT gets product information either by crawling product detail pages or directly from merchant product feeds via partnership. The two sources hand over different data. Feeds deliver checkout image URLs, brand names, merchant subtitles and "Best Price" tags 100% of the time; PDPs deliver delivery information (75% any delivery, 50% free delivery) and online availability (79%).

FieldProduct feedPDP crawl
Checkout image URL100%—
Brand100%—
Merchant subtitle100%—
"Best Price" tag100%—
Any delivery info—75%
Free delivery—50%
Online availability—79%
Top offer providers by retrieval source
Product feeds are dominated by Shopify; PDP sources are distributed across major retailers
  • Feeds · Shopifymajority
  • Feeds · other providersminority
  • PDPs · Amazon—
  • PDPs · Walmart—
  • PDPs · Best Buy—
  • PDPs · other retailers—
Relative illustration of the reported distribution.

When PDPs are important and when Product Feeds are important

Comparing 406,639 top-ranked offers with 659,868 bottom-ranked offers, product feeds win rank: practically all feed citations sit in position 1. But feeds are still a minority of supply. About 88% of all offers come from PDPs, and even merchants with an integrated feed see roughly 76% of their offers sourced from PDPs.

Retrieval source by rank position
  • Rank 1feeds 20%
  • Rank 2–3~0%
  • Rank 4+~0%
Product feedPDP crawl
Should I still care about my PDPs?
  • All merchants · PDP share~88%
  • Feed-integrated merchants · PDP share~76%

What happens when we isolate PDP-based offers

Comparing 294,913 rank-1 PDP offers with 657,282 rank-4+ PDP offers surfaces the levers that move a page up. The biggest gaps: best-price competitiveness, trust signals such as reviews and FAQs, and clear product naming that includes use cases and features.

PDP feature importance: gap between top and bottom positions
  • Best-price competitivenesslargest
  • Customer reviews
  • FAQ / Q&A present
  • Product video & images
  • Descriptive product title
  • Delivery infosmallest
Ordered by observed gap; bar lengths are illustrative of rank order.

How ChatGPT product retrieval is changing in real time

Product-feed retrieval share has risen about 15-fold from late November 2025 to June 2026, from 4.3% to roughly 20% of shopping retrievals in the last six weeks. ChatGPT is steadily reducing its reliance on traditional web search for shopping.

Share of shopping retrievals from product feeds
25%17%8%0% 4.3% ~20% Nov ’25Mar ’26Jun ’26

What this means for your brand

First, integrate a merchant product feed directly with ChatGPT so you control the metadata it sees. Then keep working the PDP, because that is still where most offers come from:

  1. Monitor competitor pricing to secure the "Best Price" designation.
  2. Build visible trust signals: customer reviews, Q&A, FAQs, video and multiple images.
  3. Write clear, descriptive product titles that include relevant use cases and features.

Methodology

30-day snapshot: approximately 1 million shopping product offers from ChatGPT sessions (406,639 top placements; 659,868 bottom placements), supplemented by scraping ~6,000 cited URLs for feature analysis. Statistical significance testing included effect-size validation. 8-month lookback: approximately 548 million unsampled shopping offers across all ChatGPT shopping-mode sessions, November 2025 through June 2026.

Allen WuResearch, Flowgen