A customer opens ChatGPT and types: “ What is the best air fryer under R$500 for a family of four? ”. In seconds, they receive three or four models, with names, price ranges, and links to buy them; no stores were visited. No Google search was performed. The decision began and almost ended there. The question that remains is simple: was your brand among the recommended ones?
ChatGPT and Perplexity have become channels for product discovery . Both already suggest specific items, compare options, and in some cases, take the user directly to checkout. For digital retail, this creates an entry point that doesn't depend on advertising or Google ranking , and which most businesses don't yet monitor.
Consumer behavior confirms this trend. According to research by Fbiz in partnership with the On The Go platform , conducted between July and August 2026 with 1.200 Brazilians, 46% of people who use artificial intelligence already resort to technology to research products or services before buying. The point of discovery has shifted, and now it responds in natural language.
On this page you will find:
AI-powered recommendations: what's already at stake?
- ChatGPT and Perplexity recommend products. Including the name, price, and link within the search results themselves, without the user having to go through a separate results page.
- Perplexity moved to direct purchase with the “Buy with Pro", which finalizes the order without leaving the conversation, and with the visual search “Snap to Shop”, which identifies a product from a photo.
- These recommendations are not advertisements. AI chooses what to quote based on public content, product data, and brand reputation on the web.
- Appearing in these responses is a new challenge., GEO (Optimization for generative engines), different from traditional SEO, although based on it.
- Traffic coming from AI converts well: according to Visibility LabsVisits from ChatGPT converted at 1,81%, compared to 1,39% for non-branded organic search.
- Those who aren't mentioned don't just lose clicks. It loses its place on the list of options that the customer will consider.
The purchase began to take place within the response.
The clearest sign came during the late 2025 shopping season. According to Adobe Analytics , traffic from AI-powered websites to stores in the United States grew 805% on Black Friday compared to the previous year . And it wasn't just volume: visitors who arrived via AI assistants were 38% more likely to complete a purchase than the average.
App data points in the same direction. Apptopia recorded a 28% increase in ChatGPT referrals for retail apps during the Black Friday weekend of 2025 compared to 2024. It's still a small slice of the total sessions, but it's growing rapidly and focuses on genuine purchase intent.
In Brazil, the habit is also taking hold. Beyond the 46% reported by Fbiz , the relevant data is qualitative. People don't type " air fryer " and choose from ten links on their own. They describe the context (budget, usage, family size) and receive a curated response . AI has taken over the screening that previously fell entirely to the consumer.
How ChatGPT and Perplexity choose the products they recommend.
The logic behind both is similar, with differences in execution.
ChatGPT
ChatGPT combines model knowledge with real-time web search and a shopping layer that displays products with images, prices, reviews, and direct links. Recommendations are not paid: OpenAI states that shopping results are chosen organically from structured product data and publicly available content. Brands that maintain complete product listings and a good reputation in reviews are more likely to appear.
Perplexity
Perplexity started as a search engine with source citation and has brought that logic to shopping. Each recommendation comes with references, making the source of the information more traceable. With " Buy with Pro ," launched in late 2024 for subscribers, users can complete their purchase without leaving the platform, and " Snap to Shop " allows users to photograph an item to find similar products for sale. The company has also opened a program for merchants to submit catalog data directly.
Both reward the same thing: clear, structured, and verifiable information about the product and the brand. There is no "optimize for AI" button. What exists is content that AI can read, understand, and trust.
Why appearing on ChatGPT and Perplexity pays off.
The drop in clicks is alarming, but the conversion rate balances the equation. A study by Visibility Labs analyzed 94 medium and large-sized stores throughout 2025 and found a conversion rate of 1,81% for traffic coming from ChatGPT, compared to 1,39% for non-branded organic search (approximately 31% higher). The average order value was lower (US$204 versus US$238), but revenue per session was higher for AI-generated traffic (US$3,65 versus US$3,30).
The explanation lies in the intention. Those who arrive via AI recommendation have already gone through a screening process: they've seen options, read a summary, and clicked on the product that made sense . It's a more qualified visit, closer to a decision. Add to that the fact that the study itself acknowledges that the influence of AI is underestimated, because many people receive the recommendation on ChatGPT and then research the brand on Google before buying, and the sale ends up being credited to organic search.
For e-commerce, it's clear that this isn't about simply replacing one channel with another. It's about ensuring a presence where a growing portion of purchasing decisions originate.
What to do to get your e-commerce site cited by ChatGPT and Perplexity
The first step is the most overlooked: being crawlable. If the robot doesn't read your product and category pages in HTML, you won't appear in the response. Content hidden behind heavy JavaScript or blocked by configuration is invisible to AI.
From there, the work aligns with good SEO practices , with adjustments to the focus.
Structure product sheets with complete and consistent data (title, description, price, availability, reviews) and apply structured product and organizational data (schema). This is what AI uses to understand and compare items.
Answer real questions in the content. Pages and articles that explain "which one to choose," "difference between," and "best for" feed exactly the type of answer that AI puts together.
Build a reputation off-site. Reviews, mentions, and presence in reliable sources weigh heavily when AI decides what to cite. It tends to recommend brands for which it finds evidence.
Maintain technical consistency. A well-defined canonical tag and consistent product information across all channels will prevent AI from being confused about which version of your page to consider.
There's no shortcut. Google has already confirmed that it ignores files like llms.txt and treats phrases designed to please robots as spam. The way to go is through real, well-structured, and trustworthy content.
Metrics that show whether you are showing up.
Tracking only organic sessions and ranking position no longer shows the complete picture. It's worth including three indicators in the dashboard.
The first is participation in AI responses: how often your brand is mentioned when someone asks about products in your category on ChatGPT and Perplexity . It functions as a voice share within generative search.
The second is AI referral traffic , isolated in analytics. Analytics tools already allow you to separate visits coming from chatgpt.com, perplexity.ai, and similar sites. This number shows whether the channel already exists for your operation.
The third is the conversion of that traffic. As market data suggests, AI-generated visits tend to convert well, but this needs to be measured in your own operation, not assumed from an external average.
Ask ChatGPT and Perplexity what a customer would ask before buying in your category and see who comes up. If the answer only shows competitors, there's content work, technical structure, and data analysis to be done to reverse that. The e-Plus SEO and GEO team works with this type of diagnosis daily , and you can find more materials about search, AI, and content here on the blog to start looking at your own numbers.
Frequently Asked Questions – ChatGPT and Perplexity
Do ChatGPT and Perplexity really recommend products to buy?
Yes. Both display product suggestions with name, price, and link within the response. ChatGPT shows purchase results with images and reviews, and Perplexity allows you to complete the order without leaving the platform using the " Buy with Pro " feature. The recommendations are organic, not paid advertisements.
What is GEO and how is it different from SEO?
GEO (Generative Engine Optimization) is the optimization of content to be cited by AI search engines such as ChatGPT, Gemini, and Perplexity. SEO aims for link ranking on Google; GEO aims to be the preferred source for assembling the answer. Both rely on clear, structured, and reliable content.
How do I get my e-commerce site to appear in AI recommendations?
Start by ensuring your pages are crawlable and viewable in HTML. Then, maintain complete product sheets with structured data, produce content that answers real buyer questions, and build reputation with reviews and mentions. There is no unique markup for AI; what works is solid SEO and verifiable information.
Is it worth investing in AI traffic if it's still small?
Yes, because it grows fast and converts well. According to Adobe, AI traffic to retail in the US increased by 805% on Black Friday 2025, and visitors coming from AI were 38% more likely to buy. It's a channel in formation, and getting there early creates an authority advantage that's hard to recover later.
