How Shoppers Use ChatGPT to Decide What to Buy (and What It Means for Brands)
Shoppers increasingly ask ChatGPT what to buy before they ever visit a store or search engine. Here is how that journey works and how consumer brands can show up in it.
A few years ago, a shopper looking for a new mattress would open a search engine, click through review sites, skim forum threads and eventually land on a few brand websites. Today, a growing number of shoppers start somewhere else. They open ChatGPT and type something like: "I'm a side sleeper with lower back pain, my partner sleeps hot, and our budget is around $1,200. What should we look at?"
That one sentence contains more buying intent than a dozen keyword searches. It names a person, a problem, a constraint and a budget. And the answer shapes which brands the shopper considers. With ChatGPT reported at around 900 million weekly users, this behavior is no longer a niche habit. This article looks at how shoppers actually use AI assistants to decide what to buy, and what consumer brands can do about it.
The shift: from keywords to conversations
Search engines trained shoppers to compress their needs into two or three words: "best running shoes," "vitamin c serum." AI assistants reverse that. People write the way they talk, and they include context they would never type into a search bar.
Compare:
| Search engine query | ChatGPT prompt |
|---|---|
| running shoes flat feet | I have flat feet and I'm training for my first half marathon. My knees hurt after 8 miles. What kind of shoes should I look for? |
| gift for dad | My dad is 68, loves fishing and cooking, and says he doesn't need anything. Ideas under $100? |
| dog food sensitive stomach | My 4-year-old lab keeps having loose stools. We switched foods twice. What should I try next? |
The ChatGPT version carries situation, history and constraints. That changes what the shopper gets back, and it changes what marketers can learn about demand.
The five stages of an AI-assisted purchase
Shopping conversations with ChatGPT tend to move through recognizable stages, though shoppers often jump between them.
1. Problem framing
The shopper describes a situation and asks what kind of solution fits. "My apartment is too dry in winter and I keep waking up with a sore throat. What helps?" At this stage, they may not know a product category exists.
2. Criteria building
The assistant explains what matters: features, tradeoffs, materials, price ranges. The shopper learns what to look for, such as "cool-mist vs warm-mist," "tank size," "noise level."
3. Shortlisting
The shopper asks for options: "What are some good quiet humidifiers for a bedroom under $80?" This is where brands get named, or do not.
4. Comparison and objection handling
"Is it worth paying more for one with a humidity sensor?" "Do these need special filters?" The shopper is stress-testing the shortlist and resolving doubts.
5. Decision and action
"Okay, which one would you pick for a small bedroom?" Then they click through to buy, often in a separate tab.
Each stage produces different questions, and each is a different opportunity for a brand. Ads that work at the problem-framing stage teach; ads at the decision stage reassure and make buying easy.
What shoppers trust, and what they don't
AI-assisted shopping has a few consistent patterns worth understanding.
Shoppers use AI to cut through noise. Many people turn to ChatGPT because review sites and search results feel cluttered with affiliate content and ads. They want a direct, reasoned answer. Brands that provide clear, verifiable product information are easier to recommend and easier to trust.
Shoppers still verify. An answer in ChatGPT is usually the start of a decision, not the end. Shoppers click through to check prices, read reviews, look at photos and confirm details. Your landing page has to do that verification work well.
Specificity wins. A product described as "great for everyone" gives an assistant nothing to match against "side sleeper who runs hot." Products with clear use cases, specific materials and honest tradeoffs fit conversational answers better.
Follow-ups reveal objections. The questions shoppers ask after the first answer, such as "does it need special filters?" or "is it hard to clean?", are a map of objections your marketing should handle.
Where ads fit in the AI shopping journey
Since February 2026, OpenAI has shown ads in ChatGPT to logged-in adult users on the Free and Go tiers in the US, and the program has expanded to more countries through the year. Ads appear below ChatGPT's answer, clearly labelled as sponsored, and do not change the answer itself. Paid tiers like Plus and Pro remain ad-free.
That placement matters for shoppers. The ad arrives after the shopper has asked a detailed question and read a response. It is not interrupting them; it is offering a next step. An ad for a quiet bedroom humidifier with a 30-day return policy, shown under a conversation about dry winter air, is useful in a way a banner ad rarely is.
Targeting works through context hints, plain-language descriptions of the situations an ad belongs in, and ChatGPT may also use what it remembers across a person's chats to choose relevant ads. That rewards brands that think in situations ("parent of a newborn in a dry apartment") rather than keywords ("humidifier"). Our guide to what ChatGPT ads are covers the format, and context hints explained shows how to write them.
Early performance signals
Data on AI-driven shopping is still early, and you should read it with that in mind. Criteo reported in February 2026 that traffic from ChatGPT converted about 1.5 times better than other channels in its data. Similarweb reported an average ChatGPT ad click-through rate around 0.68 percent, with top brands near 1.57 percent. Those figures come from specific datasets and time periods, so treat them as directional rather than as a forecast for your brand.
OpenAI has also added shopping-oriented ad formats, including product-feed and carousel formats in August 2026, which suggests the platform expects shopping to be a major use case. See our guide on ChatGPT product feed campaigns for details.
What consumer brands should change
Rewrite product information for questions, not keywords
Look at your product pages and ask: if a shopper described their situation in three sentences, would this page clearly say whether the product fits? Add use cases, who it is for, who it is not for, and honest tradeoffs. The FAQ generator can help you turn common objections into on-page answers.
Mine your own customer language
The way customers describe their problems in reviews and support tickets closely resembles how they prompt ChatGPT. Collect those phrases. They are your best source for ad copy, context hints and landing page headlines. The buyer question finder can help organize them.
Build landing pages for verifiers
AI-assisted shoppers click through to confirm. Make the page answer the follow-up questions they were about to ask: dimensions, materials, compatibility, shipping, returns. Keep claims accurate, because a shopper who just read a careful answer is quick to notice a page that overpromises.
Think in stages
Match messages to stages. A teaching message ("Cool-mist humidifiers for dry bedrooms, explained") fits problem framing. A reassurance message ("Quiet enough for a nursery, 30-day returns") fits the decision stage. Running one message across all stages leaves value on the table.
Measure the new channel properly
AI-referred shoppers often research in one session and buy in another. Use proper conversion tracking, such as the OpenAI pixel and Conversions API for ad traffic, and UTMs for links you control, and expect attribution to be imperfect. Our guide on attribution for AI channels covers practical approaches.
A worked example
Say you sell a $69 bedroom humidifier. A shopper asks ChatGPT about dry winter air and waking up with a sore throat. The answer explains humidity levels and suggests a cool-mist humidifier for a bedroom.
Below the answer, your ad appears:
- Headline:
A Quiet Cool-Mist Humidifier for Bedrooms - Body:
Runs up to 30 hours per fill with a sleep mode that dims the light. 30-day returns.
The headline is 41 characters and the body 83, within the 50 and 100 character limits. The shopper clicks and lands on a page that opens with "Quiet humidity for better winter sleep," shows the tank size, noise level, cleaning instructions and return policy, and answers "does it need filters?" before they ask. No unproven health claims, no fake urgency. Just a useful product, clearly explained, at the moment it is relevant. The runtime and features in the example would of course need to be accurate for your product.
How SecondWin helps brands meet AI shoppers
SecondWin is a done-for-you ChatGPT ads service for consumer brands. Enter your URL and it works out your product, buyer and niche. It then looks at the longest-running ads in your niche in Meta's Ad Library, since ads that run for 60+ days tend to be the ones that keep selling, and extracts the proven messages behind them along with the questions shoppers ask ChatGPT that each message answers.
It writes original ChatGPT ads as answers to those questions, checks each one against OpenAI's policies and its own rules, and, once you pick a plan and add SecondWin as an Admin on your OpenAI ad account, launches and manages the campaigns, with ad spend billed by OpenAI directly to you. See the buyer questions a free analysis finds for your store, or look at plan pricing. SecondWin is independent and not affiliated with OpenAI.
FAQ
How do shoppers use ChatGPT to buy products?
Shoppers typically describe their situation, learn what features matter, ask for options, compare them and resolve objections, then click through to a store to verify and buy. Prompts are much longer and more specific than search queries, often including budget, constraints and personal context. That makes the conversation a rich signal of what a shopper actually needs.
Do ads change ChatGPT's product recommendations?
According to OpenAI, ads do not change ChatGPT's answers. Ads appear below the response, are clearly labelled as sponsored and are visually separated from the answer. Brands that want visibility in the answer itself focus on organic approaches such as clear product information, while ads offer a paid placement next to relevant conversations.
Is traffic from ChatGPT worth more than other channels?
Early reports are encouraging but limited. Criteo reported in February 2026 that ChatGPT traffic converted about 1.5 times better than other channels in its data. Results depend heavily on your product, landing page and tracking, so treat third-party figures as directional and measure your own conversion rates and cost per acquisition before scaling.
What should brands do first to prepare for AI shoppers?
Start with your own customer language. Collect the questions and problems from reviews and support tickets, then make sure your product pages answer them clearly, including who the product is and is not for. From there, test ChatGPT ads that answer those same questions, with landing pages built for shoppers who click through to verify details.