Product selection, without the myths

How Does ChatGPT Choose Which Products to Recommend?

ChatGPT does not publish a fixed product-ranking formula. It interprets the shopper’s intent, considers available product information, and surfaces a limited set of relevant options. For Shopify merchants, being available to that process starts with eligibility and accurate product data.

Diagnose a missing product Track recommendations

Illustrative selection flow

“A waterproof commuter backpack under $120 that fits a 16-inch laptop”

Intent

commuting

Constraints

≤ $120 · 16-inch

Need

waterproof

MetroShell 24L

$109 · in stock · 16-inch sleeve

Candidate

The prompt changes which details matter. This is an explanatory mockup, not ChatGPT’s interface or algorithm.

The short answer

A product is eligible to appear; it does not earn a guaranteed position. OpenAI says relevance depends on the query and context. A stated budget makes price more important. Product descriptions, availability, reviews, and ease of use may matter when they fit the request. Not every available product is shown.

Results are selected independently and are not ads. Shopping research may also ask follow-up questions about brand, size, performance, comfort, style, or price before comparing products. OpenAI selection details · Shopping research details

A concrete example

The prompt decides which product facts become useful

For “a waterproof commuter backpack under $120 that fits a 16-inch laptop,” price, stock, laptop dimensions, and weather protection directly answer the request. A cheaper bag with no laptop-size data may still be a good product, but the available record does not establish the fit. A $139 bag fails the stated budget.

That comparison is reasoning, not a claim about hidden weights. Change the prompt to “the most durable travel backpack regardless of price,” and the useful attributes change too.

MetroShell 24L

Price: $109
Availability: In stock
Fit: 16-inch sleeve

Matches stated constraints

Studio Pack 18L

Price: $89
Availability: In stock
Fit: Laptop size missing

Match cannot be verified

StormPack 30L

Price: $139
Availability: In stock
Fit: 16-inch sleeve

Above stated budget

Illustrative comparison: it shows how explicit constraints can separate candidates, not which product ChatGPT will choose.

How Shopify products enter discovery

Catalog, feeds, and crawling are related—not interchangeable

Shopify Catalog is the primary structured product-data route for agentic storefronts. Eligible products are sent with title, description, options, images, price, availability, and other attributes. Shopify continuously updates that data.

Products can also be found through owned feeds and public pages discovered by crawling or indexing. A robots.txt block affects open-web discovery, not Catalog delivery to an activated channel. Conversely, a crawlable page does not make an ineligible product eligible for Shopify Catalog. Shopify discovery documentation

Shopify CatalogStructured product recordPrimary Shopify route

Product feedsMerchant or provider dataAdditional route

Open webCrawled product pagesSeparate route

Available information for product discovery

Catalog delivery and open-web crawling are separate. Blocking a crawler does not disable an active Shopify Catalog connection.

Confirmed

What the sources actually say

User intent includes the query, conversation context, and possibly Memory or custom instructions.
Structured first- and third-party metadata and public retail information can inform results.
Price matters more when the shopper sets a budget; availability and eligibility determine whether an offer can participate.
For multiple merchants selling the same product, OpenAI names availability, price, quality, and maker or primary-seller status.

Hypotheses to test

Useful ideas, not ranking factors

Filling a missing product attribute may improve the match for prompts that require that attribute.
Clear first-party evidence may reduce ambiguity when competitor pages describe the same use case more precisely.
Fresh price and stock data may prevent an otherwise relevant product from being represented incorrectly.
Schema, keywords, reviews, or an llms.txt file do not guarantee recommendation placement.

Competitor diagnosis

Compare evidence, not just appearances

If a competitor appears and you do not, first compare the recommendation against the exact prompt. Then inspect the facts ChatGPT could retrieve: constraint fit, product attributes, current price, availability, and cited pages.

Dropstore records repeated answers, citations, positions, mentions, and competitors for selected prompts. Pair that history with the Shopify audit to find missing or conflicting evidence. It can reveal a gap; it cannot prove the model’s private reasoning or guarantee the next result.

Start the 7-day trial

Observed gapYour productCompetitor

Budget matchYesYes

AvailabilityIn stockIn stock

Laptop sizeMissing16-inch

Waterproof proofVagueMaterial + rating

A gap suggests what to verify. It does not prove why a recommendation occurred.

Practical checklist

Why was my Shopify product not recommended?

1Write down the exact buyer prompt and every explicit constraint.
2Confirm that the product is eligible for Shopify Catalog and available to US customers.
3Check ChatGPT access under Sales channels → Agentic in Shopify admin.
4Verify title, description, category, variants, images, price, and live availability in Shopify Catalog.
5Map important metafields or custom product data into Shopify Catalog.
6Open the public product page and verify that its key claims are readable and current.
7Compare recommended products on the same facts: constraint fit, price, stock, attributes, and cited evidence.
8Repeat the same prompt set over time; never infer a rule from one result.

Need the implementation steps rather than the selection explanation? Use the Shopify ChatGPT ranking guide.

Primary sources

Source review: September 14, 2026. Eligibility and product-discovery behavior can change; verify the linked documentation before acting.