# AI Shopping Assistants: How ChatGPT and AI Mode Pick Products

URL: https://ecomchannels.com/insights/ai-shopping-assistants-chatgpt-perplexity-google-ai-mode/
Category: AI search & storefronts
Updated: 2026-10-30

> How ChatGPT, Perplexity and Google AI Mode research and recommend products, what data they rely on, and how Australian retailers can be in the shortlist.

AI shopping assistants such as ChatGPT, Perplexity and Google AI Mode take a shopper's description of what they need, run several searches, compare products on attributes, prices and reviews, and return a short list with reasons. They rely on structured product feeds, merchant data and public web pages, so the retailers that appear are the ones whose product information is complete, accurate and easy to compare. Shoppers who click through from these tools tend to arrive ready to buy.

## How the main assistants find products

**ChatGPT.** OpenAI reported more than [700 million weekly ChatGPT users in September 2025](https://openai.com/index/buy-it-in-chatgpt/), and said its product results are organic, unsponsored and ranked on relevance. Its [shopping research feature](https://help.openai.com/en/articles/12911370-using-shopping-research-in-chatgpt) asks clarifying questions (budget, brand, size, what matters most), then runs a multi-step search using merchant product data supplied through the Agentic Commerce Protocol, publicly available product information and other retail sources. It returns a buyer's guide with top picks, trade-offs, side-by-side comparisons and links to merchants that currently sell each item.

**Google AI Mode and Gemini.** Google's shopping answers draw on its Shopping Graph, which Google says holds [more than 50 billion product listings, with 2 billion refreshed every hour](https://blog.google/products-and-platforms/products/shopping/agentic-checkout-holiday-ai-shopping/). AI Mode builds comparison tables, pulls insights from reviews and shows price and inventory details. Most of that data comes from Merchant Center feeds and product structured data on retailer sites.

**Perplexity.** [Bain notes](https://www.bain.com/insights/agentic-ai-commerce-hinges-on-consumer-trust) that Perplexity added shopping features with one-click purchasing to its search results in 2024. Like the others, it combines live web retrieval with product data, and cites its sources.

The common thread: each assistant needs machine-readable facts it can compare. A product that lacks dimensions, materials or compatibility details cannot be matched to a precise request, however good the photos are.

## What shoppers are doing with them

Australians are already using AI in the buying process. The [Australia Post eCommerce Report 2026](https://auspost.com.au/content/dam/auspost_corp/media/documents/ecommerce-reports/annual-ecommerce-reports/ecommerce-report-2026.pdf) says 6 in 10 Australians now use AI. Salesforce's [2026 State of Commerce ANZ release](https://www.salesforce.com/au/news/stories/sate-of-commerce-anz-data-agentic-search-2026/) found that, between August 2025 and May 2026, product discovery through traditional search fell 15% while discovery through new channels such as AI assistants grew 38%.

Research, not autonomous buying, is where most of the activity sits. In Bain's March 2025 US survey, 72% of consumers had used AI in some form and 73% had used or would consider AI to research products or compare prices, but only 24% were comfortable letting it make a purchase today.

## Why AI-referred visitors matter

The volume is still smaller than search or email, but the quality is high. [Shopify's Q1 2026 commerce data](https://www.shopify.com/enterprise/blog/ai-search-insights) shows that more than half of AI-referred sessions start on a product page (about 20% for organic search), that these sessions convert at nearly 50% higher rates than organic search, and that average order values are 14% higher. The assistant has done the comparison work, so the visitor arrives closer to a decision.

That has two practical consequences. First, your product pages are now landing pages and must answer the final questions quickly: price, delivery time, returns, stock. Second, if the facts the assistant quoted do not match your page, you lose the sale at the last step.

## How to get into the shortlist

1. **Fill every attribute.** Size, weight, material, colour, compatibility, care, warranty, country of origin, GTIN. Assistants filter on these.
2. **Write for the question, not the keyword.** "Suits small balconies, folds flat to 9 cm, holds up to 120 kg" is far more matchable than a list of adjectives.
3. **Keep price and stock live.** Google refreshes around 2 billion listings every hour; stale prices or out-of-stock items get filtered or create mismatches.
4. **Give images that explain.** Clear main images plus dimension and in-use shots help both shoppers and visual AI features.
5. **Earn and display reviews.** Assistants summarise review themes; products with few reviews give them little to work with.
6. **Be present where assistants look.** Your own site, Merchant Center, and the marketplaces you sell on all feed discovery. See our guide to [generative engine optimisation](https://ecomchannels.com/insights/generative-engine-optimisation-geo-for-retailers/) and to [agentic commerce protocols](https://ecomchannels.com/insights/agentic-commerce-protocols-acp-ucp-explained/).

## Key figures

- ChatGPT had more than 700 million weekly users in September 2025 ([OpenAI, 2025](https://openai.com/index/buy-it-in-chatgpt/)).
- Google's Shopping Graph holds more than 50 billion product listings, with 2 billion refreshed hourly ([Google, 2025](https://blog.google/products-and-platforms/products/shopping/agentic-checkout-holiday-ai-shopping/)).
- AI-referred sessions convert at nearly 50% higher rates than organic search, with 14% higher average order values ([Shopify, 2026](https://www.shopify.com/enterprise/blog/ai-search-insights)).
- 6 in 10 Australians now use AI ([Australia Post, 2026](https://auspost.com.au/content/dam/auspost_corp/media/documents/ecommerce-reports/annual-ecommerce-reports/ecommerce-report-2026.pdf)).
- 73% of US consumers have used or would consider AI to research products, but 24% would let it buy for them today ([Bain, 2025](https://www.bain.com/insights/agentic-ai-commerce-hinges-on-consumer-trust)).

## How eComChannels helps

AI assistants reward the retailer with the most complete, consistent product data across every place they look. eComChannels is built to produce exactly that without adding headcount.

- **[AI listing optimisation](https://ecomchannels.com/products/ai-listing-optimisation/)** works from one master product record. AI already trained on each marketplace's listing rules drafts full attribute sets, question-led descriptions and titles; your team approves them; bulk updates publish to Amazon, eBay, Kogan, BIG W Market and the other channels you sell on. Complete attributes make your products matchable to the detailed requests shoppers type into assistants.
- **[Branded storefronts](https://ecomchannels.com/products/storefronts/)** are fast and SEO- and GEO-ready, and show the same price, stock and specifications as your listings, so search engines and AI assistants read consistent facts and AI-referred visitors land on a page that answers their last questions.
- **[AI graphics](https://ecomchannels.com/products/ai-graphics/)** is trained on each marketplace's image rules and builds an image set per channel; you approve once and publish everywhere, so every listing an assistant draws on has compliant, clear images.

Less manual editing, more of your range eligible to be recommended. [Book a demo or start a 30-day free trial](https://ecomchannels.com/book-a-demo/).

## FAQ

### Do AI assistants favour big brands?

They favour products they can verify and compare. Large retailers often have richer data and more reviews, but a smaller seller with complete attributes, live pricing and clear policies can be matched to specific requests that big catalogues describe poorly.

### Can I see how much traffic comes from AI assistants?

Yes. Most analytics tools record referrals from chatgpt.com, perplexity.ai and gemini.google.com. Some in-app and AI Mode visits are harder to attribute, so treat the figure as a minimum.

### Should I block AI crawlers?

Blocking them removes your products from the answers shoppers are reading. Most retailers benefit from allowing them on product and category pages while keeping account and checkout pages private.

## Sources
- Buy it in ChatGPT: Instant Checkout and the Agentic Commerce Protocol, OpenAI (2025): https://openai.com/index/buy-it-in-chatgpt/
- Using shopping research in ChatGPT, OpenAI Help Center (2026): https://help.openai.com/en/articles/12911370-using-shopping-research-in-chatgpt
- Let AI do the hard parts of your holiday shopping, Google (2025): https://blog.google/products-and-platforms/products/shopping/agentic-checkout-holiday-ai-shopping/
- AI-referred shoppers convert better and spend more: What Shopify's early data shows, Shopify (2026): https://www.shopify.com/enterprise/blog/ai-search-insights
- Agentic AI Commerce Hinges on Consumer Trust, Bain & Company (2025): https://www.bain.com/insights/agentic-ai-commerce-hinges-on-consumer-trust
- Australia Post eCommerce Report 2026, Australia Post (2026): https://auspost.com.au/content/dam/auspost_corp/media/documents/ecommerce-reports/annual-ecommerce-reports/ecommerce-report-2026.pdf
- Agentic Search Surges As 91% Of Australian Commerce Leaders Face Rising Customer Expectations, Salesforce (2026): https://www.salesforce.com/au/news/stories/sate-of-commerce-anz-data-agentic-search-2026/