ChatGPT traffic is growing, but the scale is still nowhere near Google's

Industry analyses cited in the media reveal two truths. First, ChatGPT captures a significant portion of AI traffic directed to online stores – one report estimates it accounts for over 90% of all e-commerce traffic generated by AI assistants. Second, the volume remains relatively small – it can be around 200 times smaller than Google traffic.

This means that in many online stores, AI traffic is now an "add-on" to the channel mix, not its foundation. At the same time, the growth rate is so rapid that it's worth treating it as a new user acquisition stream, requiring its own set of rules: different content, different analytics, and different trust management.

Why AI traffic often fails to sell? The most common barriers

1) Different intention: "help me choose" rather than "give me the product"

Users often come to ChatGPT for advice, comparison, and narrowing down their choices. This is a process closer to the research stage than the classic "buy now" stage. If the landing page doesn't accurately address the question (e.g., doesn't explain the differences, doesn't show options, doesn't provide quick answers about shipping and returns), the customer returns to the chat or searches further.

2) Conversion lower than in the organic channel

Studies cited by industry media show that traffic from ChatGPT can have noticeably lower conversion rates than organic search. In one study, conversion from visits from ChatGPT was approximately 13% lower on average compared to organic traffic. This doesn't necessarily mean that this channel "doesn't work"—rather, it requires adapting the shopping experience to the user's decision-making process after interacting with AI.

3) Trust and transparency of recommendations

Many users still evaluate chat recommendations through the lens of "what's the basis for recommending this?" Search is a familiar mechanism for many, and AI assistants are only just building credibility as a purchasing tool. When concerns about data security and offer transparency arise, finalizing a transaction becomes more difficult.

4) Too long and fragile checkout

AI can drive attention, but the sale ends at checkout. If a customer, after speaking with an assistant, lands on a form that's too long, poorly optimized for mobile, lacks preferred payment methods, or has unclear shipping costs, the channel will appear as "traffic without sales.".

5) Attribution and Analytics: Channel Success Can Be Invisible

Some AI traffic value isn't reflected in traditional last-click attribution. A user may have initiated their decision in chat and later purchased via brand search or email. If you don't tag your landing pages, create an "AI referrals" segment in GA4, and set up a proper funnel analysis, you'll quickly reach the erroneous conclusion that "this isn't selling at all.".

A New Direction: Chat Shopping and the Growing Role of AI Agents

To shorten the path to purchase, AI assistant vendors are testing features like "purchase without leaving the conversation." Industry discussions include the concept of checkout directly within the chat interface, which could reduce friction and increase transaction completion rates.

At the same time, the topic of AI agents, which are supposed to perform actions "on behalf of the user" (e.g., search for offers, compare, proceed to purchase), is developing. However, this trend raises tensions regarding platform control, security, and service access rules – just look at the legal disputes surrounding agent-based solutions in shopping.

Benefits and risks for online stores in Poland

What can you gain?

• New discovery stream – the user comes to you after an initial selection, often with a specific need.
• Better quality traffic in niches where advice counts: home furnishings, electronics, cosmetics, sports, B2B.
• An advantageous opportunity for brands that can clearly explain product differences and build trust through content.

What to watch out for

• Loss of relationships and data – if the role of “indirect” purchases increases, you may have less contact with the customer and less room for upselling.
• Fluctuations in traffic quality – this channel is young and user behavior is variable (sometimes large peaks without revenue).
• Reputation risk – outdated pricing, lack of availability, or a poor returns policy quickly come up in user conversations with AI.

Practical Tips: How to Make Your ChatGPT Traffic Sell Better

1) Build a landing page for the intention, not the product

AI-powered inputs are best served by pages that answer user questions: comparisons, short guides, "how to choose" sections, "for whom/when" lists, and clear differences between variants. In an online store, you can do this as:

• a category with a shopping guide above the listing,
• a problem-specific landing page (e.g. "a gift up to PLN 200", "for a small bathroom"),
• extensive FAQ on product cards.

2) Ensure consistency of information that AI easily cites

AI "likes" clear facts: parameters, dimensions, compatibility, warranty, shipping costs, processing time, returns policy. Organize this in descriptions and technical sections, and keep stock and prices up-to-date. Otherwise, the user will get a mismatch between what they read in chat and what they see on the website—and their shopping cart will crumble.

3) Shorten checkout and remove post-click friction

• Simplify the form (fewer fields, better autocomplete).
• Show shipping and returns costs upfront, not at the end.
• Empower the payments and deliveries your customers actually choose.
• Add elements of trust: reviews, certificates, clear return conditions, clear contact information.

4) Measure separately: AI referrals as a new segment

Create a segment of visits from domains and sources related to AI assistants in analytics, check:

• which landing pages receive traffic from AI,
• time on the page and scroll (whether the user is "reading" or "skipping"),
• micro conversions (adding to cart, clicking on delivery, opening FAQ),
• assisted participation in paths (not just last click).

If you want to monitor visibility in AI tools, the market already offers solutions for monitoring this type of traffic and prompts. :contentReference[oaicite:9]{index=9}

5) Strengthen your brand presence in the places that AI "reads" most often

AI assistants often rely on how-to content, reviews, comparisons, and expert materials. Maintain publications and citable knowledge: guides, specifications, and answers to frequently asked questions. This is a good time to develop a knowledge center in the digital world, for example, in the form of news and guides on your own blog. Also see our resources: digital news and knowledge base .

What's next: AI traffic will grow, but stores with the best post-click experience will win

Today's picture is clear: ChatGPT and similar tools can generate valuable discovery, but sales still depend on what happens after the site is entered. Market analysis data simultaneously demonstrates the dynamic growth of AI in the customer journey and the differences in effectiveness compared to traditional channels.
In practice, this means that "more traffic" isn't enough—the online store must be prepared for a new type of user: more demanding in terms of information, more cautious in trust, and less patient in checkout.