We Analyzed 8,520 ChatGPT Shopping Queries. Here's How to Get Recommended
ChatGPT showed products for 59.5% of 8,520 shopping queries. Intent stage decides it: 93.0% at checkout vs 2.8% after purchase. Here's how to get picked.
ChatGPT showed a product carousel for only 59.5% of the 8,520 shopping queries we analyzed, and the query's intent stage was the strongest predictor: 93.0% at Purchase Execution vs. 2.8% at Post Purchase. To get recommended, win purchase-ready queries. Deal-seeking language lifts carousels by 35 points; naming a brand narrows results to that brand. We later extended this intent-stage analysis to a larger, multi-platform sample. See our 22,335-recommendation multi-platform study covering ChatGPT, Perplexity, Google AI Mode, and Copilot for how these patterns hold up across platforms.
Correction, October 11, 2026. Three figures in this article were revised after the dataset was re-analyzed under one set of definitions: the share of citations from retail sites, the number of brands per carousel, and how far the brand-name effect generalizes. Each change is listed in Corrections and changes of the AI Shopping Visibility Study 2026, which supersedes the figures here.
We sent 8,520 real shopping queries to ChatGPT to see which ones produce product recommendations, and which get a text answer instead. The stakes keep rising. An estimated 80 million US consumers are expected to use generative AI platforms for shopping in 2026 (eMarketer, 2026). AI-referred traffic to US retail sites grew 693% year-over-year over the 2025 holiday season (Adobe Analytics via Digital Commerce 360, 2026).
Two terms matter here. Carousel trigger rate is the share of queries where ChatGPT showed products, not just text. Intent stage is where the shopper is in the buying journey, from first need to after the sale.
In this article:
- Key Takeaways
- How AI Shopping Actually Works (It Is Not What You Think)
- The 9 Stages of Shopping Intent: And Why They Determine Everything
- The 5 Signals That Make AI Show Your Products
- What Content Structure Gets Cited by AI
- Content Formats That Win in AI Search
- Mapping Content to the Customer Journey
- How to Get Your Products Recommended by AI: An Actionable Checklist
- Methodology
- Frequently Asked Questions
- Summary
Key Takeaways
- Only 59.5% of shopping queries triggered a product carousel. The other 40.5% got an informational answer with no products (AEOsome Research, 2026).
- Intent stage is the strongest predictor of whether AI shows products at all. Rates run from 93.0% at Purchase Execution to 2.8% at Post Purchase, a 90 percentage point gap.
- Naming a specific brand reduces product diversity by 68%: 5.10 distinct brands per carousel without a brand name vs. 1.64 with one.
- Deal-seeking language ("best deal," "cheapest") raises the carousel trigger rate by 35 percentage points.
- Editorial pages (publishers, review sites, listicles) are 46% of all citations, retail sites 27%, brand-owned sites 10%, marketplaces 7%, and forums 3%. Retail runs from 9% of citations at Post Purchase to 48% at Purchase Execution; brand-owned sites peak at 18% at Attribute Constrained. Independent research adds that 80% of the sources ChatGPT, Gemini, and Copilot cite don't rank in Google's top 10 for the matching prompt (Ahrefs, 2025).
These are not opinions. They are measured outcomes.
How AI Shopping Actually Works (It Is Not What You Think)
Answer Engine Optimization (AEO) means shaping your content and product data so AI answer engines pick your products. It is SEO for AI, but the signals differ. Getting recommended also depends on machine-readable product data, and our guide to Product and Offer structured data covers the JSON-LD fields, GTIN, and variant modeling AI engines read.
In traditional search, you optimize for keywords and backlinks. In AI search, the system evaluates structured product data, feed compliance, attribute completeness, and how well your content matches the intent behind a question.
The key difference: AI does not rank pages. It selects answers. A carousel is a judgment call on which products fit the need. Yours makes the cut or it doesn't.
| Dimension | Traditional SEO | AI Search (AEO) |
|---|---|---|
| What you optimize | Keywords, backlinks, page authority | Product feeds, structured data, attribute completeness |
| What the user sees | Ten blue links | Product carousels, direct answers, comparison tables |
| How ranking works | Algorithmic scoring of page signals | AI selects the best answer to a specific question |
| Conversion path | Click → site → purchase | Sometimes direct purchase with no site visit |
| Key signals | Domain authority, on-page content | Feed compliance, entity consistency, schema markup |
The 9 Stages of Shopping Intent: And Why They Determine Everything
Our research uses a 9-stage behavioral intent framework, from first awareness to after the sale. AI treats each stage very differently.
| Intent Stage | Trigger Rate | What AI Does | Example Query |
|---|---|---|---|
| Purchase Execution | 93.0% | Shows products immediately, optimized for conversion | "Where can I buy a midi dress under $100?" |
| Budget Framing | 88.8% | Surfaces price-filtered product sets | "Best dress under $150?" |
| Attribute Constrained | 81.8% | Matches specific product attributes | "I need a floral, knee-length dress" |
| Scenario Confirmation | 80.8% | Recommends products for a specific use case | "Best dress for a beach wedding?" |
| Category Exploration | 72.2% | Shows broad product selection | "What are some reliable dress options?" |
| Problem Recognition | 39.8% | Gives informational answer, occasionally shows products | "I need something to wear to an event next week" |
| Comparison | 15.3% | Provides editorial comparison, rarely shows carousel | "ASOS vs Reformation for summer dresses?" |
| Validation | 8.3% | Answers trust questions informationally | "Is Reformation worth it?" |
| Post Purchase | 2.8% | Almost never shows products | "How do I style a wrap dress?" |
The gap between Purchase Execution (93.0%) and Post Purchase (2.8%) is 90 percentage points. It is the biggest finding in the data. Intent decides whether AI shows products at all, before anything else counts.
What this means for your content strategy
Comparison and Validation queries ("Brand X vs Brand Y," "Is Brand X worth it?") almost never get a carousel. AI answers them in text. Your page may be cited, but your products won't show.
Budget Framing, Attribute Constrained, and Purchase Execution queries show products 81-93% of the time. This is where product visibility is won or lost.
The 5 Signals That Make AI Show Your Products
A few words in a query sharply change whether AI shows products.
Signals that increase product recommendations
| Signal | Effect on Carousel Trigger Rate | Example |
|---|---|---|
| Deal-seeking language | +35 percentage points | "best deal on summer dresses," "cheapest option" |
| Urgency language | +33 percentage points | "need a dress today," "fast delivery" |
| Price constraint | +30 percentage points | "dresses under $100," "budget-friendly" |
| Explicit purchase intent | +26 percentage points | "where to buy," "purchase online" |
| Multiple product attributes | +29 percentage points (at 5 attributes) | "floral, knee-length, cotton, under $150, machine washable" |
Signals that suppress product recommendations
| Signal | Effect on Carousel Trigger Rate | Why |
|---|---|---|
| Naming a specific brand | -46 percentage points | AI switches to informational mode about that brand |
| Comparison language | -49 percentage points | AI provides editorial comparison instead of product carousel |
| Problem framing | -38 percentage points | AI tries to solve the problem before recommending products |
In our analysis of 8,520 queries, deal-seeking and urgency signals each pushed carousel trigger rates above 90% on their own. Combining them added nothing. Either one was enough.
The brand paradox
This one is counter-intuitive: when a shopper names a brand, AI shows fewer products and focuses on that brand.
- Without a brand name: AI shows an average of 5.10 distinct brands per carousel
- With a brand name: AI shows an average of 1.64 distinct brands, with the named brand occupying 68% of all product slots
If your competitor's brand is the one shoppers are naming, ChatGPT shows their products almost exclusively. Share of voice in queries drives share of voice in the carousel. This narrowing also holds across engines in our May 2026 four-engine study, where a named brand shares its shelf with other brands about half the time. What does not carry over is a lower chance of getting a carousel at all: in that study, brand-named questions got product cards slightly more often than generic ones on Google AI Mode, Copilot, and ChatGPT.
What Content Structure Gets Cited by AI
Editorial pages are the largest source of AI citations in this dataset (46%), ahead of retail sites (27%), brand-owned sites (10%), marketplaces (7%), and forums (3%). Retail citations rise toward the purchase end of the journey, from 9% at Post Purchase to 48% at Purchase Execution, and brand-owned sites are cited most at Attribute Constrained (18%) and Purchase Execution (16%), where product detail matters most.
Outside research agrees. 80% of the sources ChatGPT, Gemini, and Copilot cite don't rank in Google's top 10 for the same prompt (Ahrefs, 2025). Citation and ranking measure different things.
Pattern 1: Structured comparison pages
Why it works: AI pulls from tables more easily than from prose. For "best X for Y," it looks for pages that map products to criteria.
How to execute: Use comparison tables, pros and cons per product, and a clear pick per use case. Wirecutter and CNET show up often in AI citations because their pages are built this way.
Pattern 2: Attribute-rich product pages
Why it works: More attributes, more carousels, in a near-linear line. Queries with 5 product attributes trigger carousels 87.3% of the time, vs. 58.1% for queries with zero. AI needs attributes to match products to queries.
How to execute: Give every product page full specs: material, dimensions, use cases, care, compatibility, and variants. Each added attribute lets you match more queries.
Pattern 3: Price-anchored content
Why it works: Budget Framing queries trigger carousels 88.8% of the time, the second-highest rate of any stage. Price is one of the strongest buying signals AI reads.
How to execute: Organize content by price tier: "Best [category] under $50 / $100 / $200." Show real prices, not vague value claims. AI matches them to budget queries.
Pattern 4: Scenario-specific guides
Why it works: Scenario Confirmation queries ("best dress for a beach wedding") trigger carousels 80.8% of the time. They pair a use case with intent to buy.
How to execute: Build each guide around one occasion or buyer: "Best running shoes for flat feet," "Office dresses for summer," "Gift ideas for photographers under $200." End each with clear picks.
Content Formats That Win in AI Search
Four formats match the queries that trigger products.
Format 1: "Best [Product] for [Scenario]" Pages
- When to use: Targeting Scenario Confirmation and Attribute Constrained queries (combined 82% trigger rate)
- Structure: Key criteria up front, comparison table (product, price, best for, key feature), product sections with pros/cons, a verdict per use case
- Example heading: "Best Summer Dresses for Beach Weddings: 12 Options by Budget and Style"
Format 2: Price-Tier Buying Guides
- When to use: Targeting Budget Framing queries (88.8% trigger rate)
- Structure: Price tier sections ($0-50, $50-100, $100-200, $200+), 3-5 products per tier with specifications, a comparison table and "best value" callout per tier
- Example heading: "The Best Midi Dresses at Every Price Point (2026 Guide)"
Format 3: Product Attribute Matrices
- When to use: Targeting Attribute Constrained queries (81.8% trigger rate)
- Structure: Sortable filter table with a column per major attribute (material, size range, price, rating, use case); each row links to a full review
- Example heading: "Women's Dresses Compared: Material, Fit, Price, and Occasion Guide"
Format 4: Category Decision Guides
- When to use: Targeting Category Exploration queries (72.2% trigger rate)
- Structure: "What to look for" section, decision tree, top picks by buyer profile (beginners, budget-conscious, quality-focused), FAQ section
- Example heading: "How to Choose the Right Dress: A Complete Decision Guide"
Mapping Content to the Customer Journey
Each phase of the funnel needs a different approach.
| Phase | Intent Stages | Carousel Rate | Content Strategy | Priority |
|---|---|---|---|---|
| Awareness | Problem Recognition | 39.8% | Educational blog posts, problem-solution content | Medium: builds authority for later-stage citations |
| Consideration | Category Exploration, Budget Framing, Attribute Constrained | 72-89% | Buying guides, comparison tables, price-tier pages | High: this is where product visibility peaks |
| Decision | Comparison, Validation, Scenario Confirmation | 8-81% | Reviews, brand comparisons, scenario-specific guides | Mixed: Scenario Confirmation is high-value; Comparison/Validation are editorial |
| Purchase | Purchase Execution | 93.0% | Product pages, "where to buy" content, pricing pages | Highest: 93% carousel rate, direct conversion |
| Retention | Post Purchase | 2.8% | Care guides, styling tips, FAQs | Low for product visibility; high for customer retention |
For most eCommerce merchants, the Consideration phase is the highest-leverage opportunity. It pairs high carousel trigger rates (72-89%) with broad reach. It catches shoppers who haven't picked a brand yet.
How to Get Your Products Recommended by AI: An Actionable Checklist
- Audit your product feed for completeness: ensure every SKU has a GTIN, full attribute set, accurate pricing, high-quality images, and structured schema markup (JSON-LD product schema at minimum)
- Add comparison tables to your top category pages: AI extracts tables more easily than paragraphs; include price, rating, key features, and "best for" columns
- Create price-tier buying guides: Budget Framing queries trigger carousels 88.8% of the time; organize picks by price bracket
- Build scenario-specific product pages: target "best X for Y" queries by mapping products to use cases, occasions, and buyer profiles
- Include pros and cons for every recommendation: AI favors balanced evaluations over promotion; cover multiple products, not just your own
- Maximize product attributes: each added attribute raises carousel trigger probability; aim for 3+ structured attributes per product
- Add FAQ sections in natural language: phrase questions the way a shopper asks an AI assistant; answer in 2-4 direct sentences
- Add structured data markup: use Product, FAQ, and HowTo schema where it applies; AI systems read these signals when building answers
- Monitor your AI visibility by intent stage: track which query types surface your products; prioritize Budget Framing, Attribute Constrained, and Purchase Execution
- Update pricing and availability quarterly: stale prices or out-of-stock products reduce recommendation likelihood
Methodology
This analysis is based on 8,520 shopping intent queries run against ChatGPT (gpt-5-mini), covering all 9 intent stages and 5 query styles in the Women's Fashion category, US market. Data was collected in February 2026. We measured carousel trigger rates, product and brand diversity, citations, and response behavior.
Query styles varied phrasing within each stage: direct question, conversational, comparative, constraint-first, and scenario-first. Brand diversity is the average count of distinct brands across triggered carousels for a given condition, with brands resolved from product URLs and titles (an earlier version counted a truncated brand field, which inflated the counts). Full study: the AI Shopping Visibility Study 2026 (version 2026-10) re-analyzes this dataset together with a four-engine, seven-category study and is the reference for every figure in this article.
The 9-stage framework is AEOsome's own classification, built from how the 8,520 queries clustered by intent. The stages apply across eCommerce categories. The percentages reflect this dataset (Women's Fashion, US, ChatGPT), not fixed rules for every category.
Frequently Asked Questions
How does ChatGPT choose which products to recommend?
ChatGPT weighs feed data, attributes, pricing, ratings, and relevance. In our analysis of 8,520 shopping queries, the strongest predictor was intent stage. Purchase Execution queries triggered product carousels 93% of the time; Post Purchase queries, less than 3%. Products with full attributes, accurate prices, and schema markup are far more likely to appear.
What is Answer Engine Optimization (AEO)?
AEO means shaping your content and product data so AI answer engines (ChatGPT, Perplexity, Claude, Gemini) recommend your products. SEO focuses on rankings and backlinks. AEO focuses on feed compliance, structured data, attribute completeness, and entity consistency.
How can eCommerce stores optimize for AI search?
Start with product feed completeness: GTINs, full attribute sets, accurate pricing, and JSON-LD product schema. Then build price-tier guides, scenario pages, and comparison tables for the high-trigger stages. In our research, queries with 5 product attributes triggered carousels 87.3% of the time vs. 58.1% with zero attributes.
What is the difference between SEO and AEO for eCommerce?
SEO earns rankings through keywords, backlinks, and authority. AEO earns AI recommendations through feed compliance, structured data, and extractable content. SEO produces ranked links. AEO produces product carousels, direct answers, and sometimes purchases made by the AI. Both matter.
What content format gets cited most by AI for shopping queries?
Comparison tables with price, features, pros/cons, and "best for" columns are the easiest format for AI to extract. In our data, editorial pages are 46% of AI citations, retail sites 27%, and brand-owned sites 10%; retail citations peak at Purchase Execution (48%) and brand-owned citations at Attribute Constrained (18%).
Does naming a brand in an AI shopping query help or hurt product visibility?
It narrows the response sharply. In our dataset, branded queries had a 21.1% carousel trigger rate vs. 66.6% for unbranded ones, a 46 percentage point drop. When a carousel did appear, brand diversity fell from 5.10 distinct brands to 1.64, with the named brand taking 68% of product slots. Unbranded queries are where the widest opportunity sits. The carousel-rate drop is specific to this ChatGPT dataset; in the May 2026 four-engine study, brand-named questions were slightly more likely than generic ones to get product cards.
Summary
- Primary finding: Intent stage is the strongest predictor of AI product recommendations, with a 90 percentage point gap between the best- and worst-performing stages
- Best content type: Comparison pages with price tiers, attribute tables, and scenario picks
- Key triggers: "best deal," "under $X," "buy," "today," "best [product] for [scenario]"
- Required structure: Comparison table + pros/cons + FAQ section + structured schema markup
- Quick win: Add comparison tables with price, rating, and "best for" columns to your top 10 category pages
About the author. Vijaya Kumar Channalli is a serial SaaS founder and the founder of AEOsome (Tartu, Estonia), which helps eCommerce brands get their products recommended when shoppers ask AI what to buy. He designed the method and the nine-stage intent framework behind this 8,520-query study. He writes about AI shopping visibility, product feeds, and how AEO differs from SEO. More at vijay.cc. Read how we write, source and correct posts in our editorial policy.
Sources: AEOsome Research, 2026 (8,520-query ChatGPT shopping-intent study, Women's Fashion category, US, gpt-5-mini, February 2026); eMarketer, The State of AI Shopping Assistants (June 2026); Digital Commerce 360 / Adobe Analytics, generative AI holiday shopping traffic (January 2026); Ahrefs, AI search overlap study (2025); Search Engine Land, ChatGPT shopping carousel source analysis (March 2026).
