How People Shop with ChatGPT—and What Merchants Need to Know

A shopper asks ChatGPT for a carry-on suitcase under $200. The first answer gives them somewhere to start. The next questions—about cabin size, wheels, warranty and delivery—help them decide which option is actually worth buying.

For a merchant, each question reveals what the offer needs to explain. Being mentioned is useful. Helping the customer resolve the next doubt is what keeps the decision moving.

A Buying Question Starts a Decision Journey

A shopper who asks for “the best carry-on” may still be working out what best means. A lower weight, a stronger warranty or a size suited to a particular airline can lead to different choices.

ChatGPT’s product discovery experience supports browsing, refining and comparing options. For merchants, that makes the conversation a useful way to understand the information behind a buying decision.

The five stages below organize those decisions: understanding the category, finding suitable options, comparing trade-offs, checking the deal and resolving final doubts. A customer may skip a stage or return to an earlier question. The purpose is to see what would help them move forward.

Five Stages—and What Merchants Need to Provide

1. Research: “What Do I Actually Need?”

Before choosing a suitcase, a first-time buyer may need to understand cabin dimensions, capacity, wheel construction and the difference between a hard and soft shell. They are establishing the criteria they will use later.

A useful merchant explanation connects each feature to a situation. Exterior dimensions matter for airline limits. Empty weight affects how much the traveler can pack. Wheel construction matters when the bag is used frequently.

Give the buyer enough context to form a useful requirement. A feature list leaves that work to them.

2. Shortlisting: “Which Products Fit My Situation?”

The request becomes more specific: “I need a carry-on under $200 for frequent international travel.” The shopper may add an airline, a weight limit or a need for a laptop compartment.

Now the question is which models meet those conditions together. Publish exterior dimensions including wheels and handles, weight, available variants and relevant use cases. Keep the exact model clear so an assistant can compare like with like.

A “carry-on” label alone cannot establish airline fit. The bag’s dimensions and the airline’s current allowance both matter.

3. Comparison: “Is the Premium Version Worth It?”

Two models may fit the basic requirements but make different trade-offs. One could cost less; another could have replaceable wheels or different warranty coverage.

Explain what the extra money changes for the traveler. Show comparable measurements and the applicable terms together. A frequent traveler can then judge whether the difference matters enough to justify the price.

That is the information a recommendation needs to answer “better for me,” rather than simply “more expensive.”

4. Price and Timing: “Should I Buy This Now?”

A suitable bag that arrives after the trip does not solve the problem. The customer needs the current offer for the selected variant: price, any promotion conditions, delivery cost and an applicable arrival estimate.

Put those details where the customer chooses the item. A starting price or a general shipping promise may leave the actual purchase unresolved.

The merchant page should show the same offer discussed in the recommendation. A different variant, an expired discount or an unexpected delivery charge can reopen a decision the shopper thought was settled.

5. Validation: “Am I Missing Anything?”

Before paying, the customer may ask what the warranty excludes, how returns work or whether reviews support a durability claim. Answering these questions helps the customer judge the risk of choosing the wrong option.

Make the supporting evidence easy to inspect: the relevant warranty, return conditions, product documentation and reviews that describe actual use. State exclusions close to the promise they qualify.

Do not imply that a retailer’s return policy covers airline rejection unless the policy says so. A clear answer helps the buyer decide whether the remaining risk is acceptable.

For the suitcase example, the merchant information maps to the stages like this:

How One Conversation Moves Through the Stages

The stages become easier to use when you follow one customer’s changing requirements:

1. “What makes a suitcase suitable for frequent international travel?” The shopper starts with research. Explain size, weight, construction and repair options before pushing a model.
2. “Which carry-ons under $200 fit my airline’s allowance?” The shopper can now narrow the range. The exact dimensions and current allowance decide which models belong in the shortlist.
3. “Is the more expensive one worth it for the wheels and warranty?” The comparison moves from matching specifications to weighing a trade-off. Show what differs and who would benefit.
4. “Can it arrive before Friday?” Timing can eliminate the preferred model. The assistant may need to return to the shortlist using the delivery requirement.
5. “What happens if it is the wrong size when it arrives?” The customer needs the applicable return conditions before committing.

The questions do not have to follow a fixed order. A new requirement can change an earlier answer. The merchant’s job is to keep the relevant facts connected so the customer can revise the choice without starting over.

A shorter shortlist can therefore lead to deeper questions. The value is a more informed decision, not simply fewer clicks.

Why a Good Recommendation Still Loses the Sale

By the time the customer reaches your page, they may have a specific reason to consider the product. The page needs to carry that reason forward.

If the recommendation described a lightweight bag with replaceable wheels, the customer should find the exact model and supporting information. A generic homepage makes them repeat the search. A product page that omits the wheel details leaves the recommendation unresolved.

Look for the point where the promise and the destination stop agreeing:

• The item changes. The link opens another variant or a broad category.
• The deal changes. The quoted price excludes a condition, or delivery misses the required date.
• The evidence disappears. A claimed benefit is not supported on the page.
• The remaining risk is unresolved. Warranty, compatibility or return questions remain unanswered.

Review the recommendation and the landing experience together. Even when the link works, an unanswered buying question can still stop the purchase.

Turn Buying Questions into a Stronger Offer

Start with one product and a question your team hears before a sale. Identify the decision it changes, then put the answer beside the relevant product, variant or purchase condition.

For a suitcase, that might mean dimensions and delivery. For software, it may be an integration and a plan limit. For a hotel, it could be a late shuttle and cancellation terms. The details change by category; the task stays the same: help the buyer decide whether the offer fits.

A specification becomes useful when the customer can understand what it means for their own situation. Support that explanation with current terms and a next action that continues the same choice.

These questions also reveal where your business has a reason to win a customer. A traveler who needs a late shuttle has expressed something more useful than a broad interest in hotels. A merchant with the right service has an offer worth bringing into that decision.

AON—Agent Offer Network—connects those offers with relevant requests inside compatible AI products. Highly structured commercial information helps applications understand and compare the merchant, terms and next action. Matching connects the offer to the request; attribution links qualifying outcomes to the referral.

For merchants, this creates a way to reach customers while they are choosing. You can fund a qualifying action or sale through CPA/CPS terms and assess the business it brings. AON provides this commercial route alongside discovery through independent assistants such as ChatGPT.

Bring one offer, its strongest buying use case and target market to discuss reaching relevant demand through AON. The merchant preparation guide can then help your team organize the work for that opportunity.

Measure Whether the Journey Creates Better Customers

Measure the part of the journey you are trying to improve. Product mentions, referred visits and valuable customers answer different questions.

• Representation: Does the public information help an independent assistant explain the product and its fit? Use a consistent set of buying questions to identify missing or conflicting facts.
• The landing journey: Do referred visitors reach the intended variant, find the answer they need and continue? Review abandonment and recurring customer questions alongside conversion rate.
• Customer value: Compare qualified leads or sales, acquisition cost, margin, returns and repeat purchases. A recommendation that produces unsuitable orders can add cost even when conversion rises.

Adobe’s October 2026 U.S. retail report found 71% year-over-year growth in AI-referred traffic in September. Its analysis covers more than 1 trillion visits to U.S. retail sites. That provides market context; your own customer results should guide the investment.

A referral report captures visits with an identifiable source. It does not capture every conversation that influenced a purchase. Keep wider organic influence separate from directly attributed commercial outcomes. For AON campaigns, use attributed qualifying outcomes to assess the customer value and acquisition cost of each offer.

The strongest offer keeps answering the customer’s questions as the decision becomes more specific. Begin with the question that holds up a purchase, connect it to a useful next step and judge the improvement by the customers it helps your business acquire.

FAQ

How should I structure product information for ChatGPT shoppers?

Organize product information around comparison, discovery and verification. State who the product fits, where it falls short, what it costs and which evidence supports the claims. A shopper comparing two models should be able to find the difference without contacting your team.

What can stop an AI-referred visitor from completing a purchase?

A mismatch between the recommendation and the merchant destination can stop an AI-referred visitor from buying. The suggested variant may be unavailable, or the displayed price, delivery date or return condition may differ from what the shopper expected. Follow the actual referral journey and compare the promise with the purchase option. Then use customer questions and abandonment points to identify what still needs explaining.

Does ChatGPT shopping matter for my category?

ChatGPT shopping matters most when your customers use it to research, compare or validate purchases in your category. Select a few buying tasks based on real customer questions, inspect the answers and review identifiable referred visits. Look for unresolved fit, price or purchase-condition questions your business can answer. Use those observations to decide where to invest; a category appearing in one AI answer does not establish a reliable acquisition channel.

How can merchants use ChatGPT shopping questions to improve product content?

Group them by the decision they change: compatibility, total cost, delivery, maintenance or returns. Put each answer beside the relevant product fact and support it with an approved source. Recheck whether a customer can now resolve the question without contacting your team. Improve the explanation before adding another generic FAQ block.

Which parts of the ChatGPT shopping journey should merchants improve first?

Begin where an unresolved question has the clearest effect on a purchase. If buyers cannot identify a suitable variant, improve shortlisting and fit information. If they arrive with a choice but abandon after checking delivery or returns, work on validation and the landing journey. Use observed customer problems to choose the stage, rather than rewriting the whole catalog at once.

What should I measure to see whether ChatGPT is influencing sales?

Track identifiable AI-referred visits, qualified customer actions and sales, while keeping broader influence separate from directly attributed revenue. Compare conversion and customer value with other channels using enough observations to support a decision. AON offers a complementary commercial route by matching structured merchant offers to requests in compatible AI products and attributing qualifying outcomes. Review those results separately from organic ChatGPT referrals so the sources and economics remain clear.