Can Electronics-Retailer EC Sites Surface Among ChatGPT's Product-Card Purchase Options? A Survey of 104 Purchase Options Across Five Major Appliance Categories

Summary

Stellagent, Inc. examined the product cards and product detail panels shown in ChatGPT's shopping consultations to see how often electronics-retailer EC sites appear as purchase options, across five categories: refrigerators, washing machines, air conditioners, TVs, and vacuum cleaners. On August 4, 2026, we observed 15 prompts in the ChatGPT web UI temporary chat, clicking the displayed product cards to record the purchase options, prices, and stock/delivery information shown in the detail panels. The three prompts used in this survey are reproduced in full in the appendix at the end of this article.

Metric	Result
Target AI	ChatGPT web UI (temporary chat)
Survey categories	Refrigerators, washing machines, air conditioners, TVs, vacuum cleaners
Scenarios	5 categories × 3 personas = 15 prompts
Product cards	42
Products with confirmed purchase options	38
Products without confirmed purchase options	4
Purchase options	104
Electronics-retailer purchase options	23 (22.1%)
EC-marketplace purchase options	47 (45.2%)
Purchase options with a confirmed price	104
Purchase options with confirmed stock/delivery info	104

Of ChatGPT's 42 product cards, 38 showed purchase options, and the detail panels displayed 104 purchase options in total. Of these, electronics-retailer EC sites were 23 (22.1%), below EC marketplaces at 47 (45.2%); the most common purchase-option category was EC marketplaces. Among named purchase options, Amazon's official site was the most frequent at 25, and among electronics-retailer EC sites, Yamada Denki (7), Joshin (6), and edion.com and biccamera.com (3 each) ranked at the top.

You may reproduce the figure above in news coverage, articles, and other materials as-is, provided you credit "Survey by Stellagent, Inc." and link to this page.

The share of electronics-retailer EC sites varied widely by category: refrigerators were 9 of 25 (36.0%), while vacuum cleaners were 2 of 22 (9.1%). In addition, all 104 recorded purchase options showed price and stock/delivery information.

Background

Generative-AI shopping consultations are expanding from the traditional experience of "asking for recommended products in text" to one accompanied by product cards, product details, and purchase links. OpenAI's help documentation explains that ChatGPT Search's shopping feature may display product candidates, product details, and links to review details or make a purchase in response to questions with buying intent.

For electronics retailers, this change cannot be captured only by "whether the store name is recommended in the AI's answer text." After a product itself is displayed in the AI, which EC site the user can navigate to, and which sellers line up alongside price, stock, and delivery conditions, become a new exposure surface for EC referrals. This survey therefore focuses on ChatGPT's product cards and detail panels and examines whether electronics-retailer EC sites appear as purchase options in major appliance categories.

Scenarios Used

The target categories are five: refrigerators, washing machines, air conditioners, TVs, and vacuum cleaners. For each category, we ran scenarios for three personas (P1–P3) with different buyer profiles. The full text of the prompts actually entered for each scenario is provided in the appendix at the end of this article.

ID	Scenario	Condition
P1	Single person, new life	Asks for a product that is not too expensive and easy to place in a one-room apartment
P2	Newlyweds, two-person household	Asks for a product that lasts and is easy to use
P3	Four-person family with children, replacement	Asks for a product suited to family use, such as bulk buying/bulk washing

Electronics-Retailer EC Sites Were 23 of 104 Purchase Options

Within the scope of this survey, of the 104 purchase options displayed in ChatGPT's product detail panels, electronics-retailer EC sites were 23 (22.1%). EC marketplaces, at 47 (45.2%), were the most common category among the purchase options.

Purchase-option category	Purchase options	Share
EC marketplaces	47	45.2%
Specialty EC/other	24	23.1%
Electronics retailers	23	22.1%
Manufacturer official	7	6.7%
Used/secondary	3	2.9%
Total	104	100.0%

While electronics-retailer EC sites appeared a certain number of times, EC marketplaces, specialty EC, manufacturer official stores, and used/secondary sellers lined up as purchase options within the same product detail panels. Separately from whether a store name is mentioned in a product recommendation, on the exposure surface of a product card's purchase options, electronics-retailer EC sites came in below EC marketplaces.

Amazon's Official Site Was the Most Frequent Named Purchase Option

Among named purchase options, Amazon's official site was the most frequent at 25. Among electronics-retailer EC sites, we confirmed Yamada Denki (7), Joshin (6), edion.com (3), biccamera.com (3), K's Denki Online Shop (2), Kojima.net (1), and Sofmap.com (1).

Rank	Purchase-option name	Count	Category
1	Amazon official site	25	EC marketplace
2	Yamada Denki	7	Electronics retailer
3	Joshin	6	Electronics retailer
3	Xprice	6	Specialty EC/other
5	yahoo.co.jp	4	EC marketplace
6	edion.com	3	Electronics retailer
6	Costco Online	3	Specialty EC/other
6	Panasonic Store Plus	3	Manufacturer official
6	biccamera.com	3	Electronics retailer
10	Second Street	2	Used/secondary
10	K's Denki Online Shop	2	Electronics retailer
10	Askul	2	Specialty EC/other

In this count-based ranking of purchase-option names, options with the same count share the same rank, and the next rank is shifted down by the number of tied options. The counts are the number of occurrences of purchase-option names displayed in ChatGPT's product detail panels, and do not indicate any seller's sales performance or price competitiveness.

By Category, Refrigerators Had the Highest Electronics-Retailer Share

The share of electronics-retailer purchase options varied by category. For refrigerators, 9 of 25 (36.0%) were electronics retailers, whereas for vacuum cleaners only 2 of 22 (9.1%) were.

Category	Product cards	Purchase options	Electronics-retailer purchase options	Electronics-retailer share
Refrigerators	9	25	9	36.0%
TVs	7	19	5	26.3%
Washing machines	9	24	5	20.8%
Air conditioners	8	14	2	14.3%
Vacuum cleaners	9	22	2	9.1%

You may reproduce the figure above in news coverage, articles, and other materials as-is, provided you credit "Survey by Stellagent, Inc." and link to this page.

Within the scope of this survey, categories where size and delivery/installation considerations are more relevant may be more likely to include electronics-retailer EC sites among their purchase options. However, each category was observed with only three prompts, so additional research is needed for generalization.

Price, Stock, and Delivery Information Was Confirmed for All 104 Options

For the 104 purchase options recorded this time, price and stock/delivery information was confirmed for every option. This shows that ChatGPT's product detail panels can function not merely as recommendation text but as a purchase-consideration surface close to seller comparison.

That said, the displayed prices and stock/delivery conditions are results of UI observation at the time of the survey and do not guarantee actual selling conditions or the latest information. OpenAI's help documentation also notes that there may be a delay before price or delivery-condition updates are reflected in the display.

Implications for Businesses

Electronics retailers need to broaden their generative-AI measures beyond "whether the store name appears in the AI's answer text" to include "whether their own EC site is displayed among a product card's purchase options." In this survey, while electronics-retailer EC sites appeared a certain number of times, EC marketplaces, specialty EC, manufacturer official stores, and used/secondary sellers also lined up as purchase options within the same product detail panels.

In particular, how information such as product data, price, stock, delivery, installation, warranty, and in-store pickup is taken into the AI's product and seller metadata and displayed may affect future EC referrals. It is important for electronics retailers to monitor their product-card exposure on ChatGPT for each major category, and to check the state of their official EC's product information, structured data, stock/delivery information, and seller feeds.

Also, because purchase options appear differently by product category, it is necessary to monitor separately large/installation-type categories such as refrigerators, washing machines, and air conditioners, and categories that tend to be completed more readily via EC, such as vacuum cleaners and TVs.

Survey Method (Details for Reproduction)

This survey is an AI-answer experiment and UI observation targeting the ChatGPT web UI. So that readers can reproduce and verify it under the same conditions, we disclose the operator, subjects, selection criteria, target AI and conditions, control of contextual influence, items recorded, extraction/classification/aggregation methods, deviations/exclusions, external sources, and the handling of ethics and privacy.

The survey objective is to confirm how much electronics-retailer EC sites are exposed among the product cards, product detail panels, and purchase options shown in ChatGPT's shopping consultations. The survey is conducted by Stellagent, Inc., and the operator is Codex. The subjects are ChatGPT web UI product cards, product detail panels, and purchase options, and the survey date is August 4, 2026. The selection criterion was a ChatGPT web UI from which product cards and detail-panel purchase options could be observed in this survey environment as of August 4, 2026. The target categories were selected as major appliances involving both EC purchase and store/installation/warranty consultation: refrigerators, washing machines, air conditioners, TVs, and vacuum cleaners. The three personas are: single person starting a new life; newlyweds in a two-person household; and a four-person family with children making a replacement purchase.

Target AI and Conditions

The AI service is the ChatGPT web UI, and the mode is temporary chat. The model's internal version and subscription plan are not included in this survey's observation records. For web search and shopping display, we observed the product cards and product detail panels within the ChatGPT web UI. The observation unit was 5 categories × 3 personas = 15 prompts, each entered once. There were 0 invalid observations; products for which a product card or purchase option could not be confirmed were not deleted and were recorded as having no purchase options.

Control of Contextual Influence

Each prompt was entered in the ChatGPT web UI temporary chat, designed to suppress the influence of past conversations.

Items Recorded

We entered each prompt in the ChatGPT web UI temporary chat, clicked the displayed product cards, and recorded the purchase options, prices, and stock/delivery information shown in the product detail panels. For records, we saved screenshots, DOM/text logs, and JSON.

Extraction, Classification, and Aggregation

For each purchase option, we opened the product detail panel by clicking a product card and counted as one each option that displayed a seller name, price, and stock/delivery information. Product cards were counted as one each for the product-candidate UI displayed within ChatGPT's answer. In aggregation, we counted purchase options individually and classified seller names as "electronics retailer," "EC marketplace," "manufacturer official," "specialty EC/other," or "used/secondary."

Classification followed these criteria. We classified Yamada Denki, Joshin, edion.com, biccamera.com, K's Denki Online Shop, Kojima.net, and Sofmap.com as electronics retailers. We classified Amazon's official site, Rakuten Ichiba, Yahoo! Shopping, and the like as EC marketplaces; when an individual store name within an EC marketplace was displayed, we classified it as an EC marketplace because the display surface is an EC marketplace. Manufacturer official included options displayed as a manufacturer's official store or a manufacturer-affiliated official sales surface. Specialty EC/other included specialty EC, general mail-order, and other sales surfaces not classified as electronics retailer, EC marketplace, manufacturer official, or used/secondary. Used/secondary included options displayed as used sales, reuse, or secondary distribution. Purchase-option names were classified based on the display names in ChatGPT's product detail panels.

Deviations and Exclusions

All 15 prompts were treated as valid observations. Products for which a product card or purchase option could not be confirmed were not deleted and were recorded as having no purchase options. In this observation, there were 4 products for which a purchase option could not be confirmed.

External Sources and Verification Method

As an external source, we used the OpenAI Help Center article "Shopping with ChatGPT Search" (checked on August 5, 2026). We used it to confirm that ChatGPT Search's shopping feature may display product candidates, product details, purchase or detail-review links, and seller candidates; that product results are not advertisements; and that seller candidates are displayed based on product and seller metadata and the like.

Reference	Provider	URL	Purpose
ChatGPT Search shopping feature	OpenAI Help Center	Source	Background on the observed UI and explanation of survey limitations

Ethics and Privacy

This survey observes the product cards and product detail panels displayed in ChatGPT's public UI; it is an AI-answer experiment and UI observation, not a survey of human subjects, so no personally identifying information was collected. Each prompt was entered in a temporary chat to suppress the influence of past conversations.

Notes on the Survey

• This survey is an exploratory study targeting the ChatGPT web UI as of August 4, 2026, and is not intended for statistical generalization.
• The display of product cards, purchase options, prices, stock, and delivery may vary with the date and time, UI state, account state, region, product data, and sellers' update status.
• Purchase options include listings that appear to be weakly related to the product and sales surfaces on overseas domains; this survey does not verify the authenticity, existence, or domestic-delivery availability of each option.
• This survey observed the purchase options displayed in ChatGPT's product detail panels and did not verify the consistency of stock, price, delivery conditions, or installation conditions after navigating to each EC site.
• Purchase-option names were classified based on the display names in ChatGPT's product detail panels. Seller names, in-marketplace store names, used/secondary displays, and the like depend on the display at the time of the survey.
• This survey does not evaluate the superiority of any seller or product.
• Other AI services such as Gemini, Copilot, and Perplexity are not included in the main analysis of this survey.
• Product and service names in this report are trademarks or registered trademarks of their respective companies.

Revision History

• August 6, 2026: First edition published

Prompts Used

For each category, we ran the following three prompts. In {product}, we inserted one of: refrigerator, washing machine, air conditioner, TV, or vacuum cleaner. The prompts were entered in Japanese; the English translations are shown here.

Single person, new life

I'm starting to live on my own and want to buy a . Do you have any recommendations that aren't too expensive and are easy to place in a one-room apartment?

Newlyweds, two-person household

I'm getting married and starting a two-person household, and I want to buy a . Do you have any recommendations that last a long time and are easy to use?

Four-person family with children, replacement

I'm in a four-person family with children and want to replace our . Do you have any recommendations suited to family use, such as bulk buying or bulk washing?