The traffic is smaller. That does not make it unimportant.

If ChatGPT, Perplexity, Gemini, or an AI-generated Google result sends a handful of visitors to your site, the sensible reaction is not to rebuild your website around them. The sensible reaction is to inspect what those visitors expected to find.

AI-referred traffic is still a small channel for many businesses. Ahrefs found that 63% of the sites in its study received at least one visitor from an AI chatbot, but that does not mean AI was a large share of their visits. Trust Insights reported 1,066 AI-referral sessions across six months, with ChatGPT accounting for 706 of them. Those are useful signals, not proof that a new acquisition channel deserves a large budget.

The case for a dedicated landing page is therefore narrower than many AI-search articles suggest. Build one when the referral represents a recognisable buying situation and your current page makes the visitor work to connect your offer to that situation. Do not build one simply to place phrases such as “AI search” or “ChatGPT visitors” on a page.

Google’s current guidance supports that restraint. Pages eligible for AI Overviews and AI Mode must still be indexed and eligible for ordinary Search. Google says there are no special technical requirements or separate AI markup that guarantees inclusion. Its advice is to create useful, original pages for people, then measure what happens after the click. (developers.google.com)

A dedicated page or a stronger existing page?

AI referrals often arrive after part of the selling has already happened.

The important distinction is not “AI traffic versus Google traffic.” It is whether the visitor arrives for discovery or confirmation.

A person asking an AI tool to find a bookkeeping firm for a construction company, compare two scheduling systems, or recommend a local photographer for a specific type of event has already supplied context. The AI may have narrowed the field, summarized the options, and attached your business to a reason for consideration. The click is often a request for verification: Do you really do this? Can I trust you? What happens next?

That helps explain why the available numbers disagree. Similarweb reported that AI-referred visits to technology platforms converted at 11.4% in June 2025, compared with 9.3% for paid search and 5.3% for organic search. Adobe’s retail data showed the opposite pattern: in July 2025, generative-AI traffic was 23% less likely to convert than non-AI traffic, although the gap had narrowed from 49% in January. Google says clicks from results pages containing AI Overviews tend to be higher quality, with visitors more likely to spend time on the site, but it does not claim that every AI source or industry converts better. (similarweb.com)

These findings are not interchangeable. Similarweb’s figure concerns technology platforms. Adobe’s concerns retail purchases, where AI can influence research well before someone is ready to buy. A service business may see a different pattern again. The practical conclusion is stronger than any universal benchmark: measure AI referrals by landing page, intent, assisted conversion, and lead quality. Do not assume a high conversion rate from a tiny sample is a durable channel advantage.

What practitioners are seeing on the ground

Practitioner reports point to a recurring pattern: AI traffic is often too small to justify a separate acquisition program, yet specific enough to expose weaknesses in the website.

Refine Labs reported that its AI-referred sessions grew roughly five to six times from January to December 2025 after changing its content structure and technical signals. The company notes that it is reporting directional growth rather than publishing exact session counts. That distinction matters. Growth from a very small base can look dramatic while remaining commercially modest. Still, the report shows why the channel deserves observation rather than dismissal.

Seer Interactive’s analysis of one client found that visitors from ChatGPT and other AI sources converted at a higher rate than Google organic traffic. Again, this is a client case, not a market average. Trust Insights took a more operational approach: it built separate analytics segments for ChatGPT, Perplexity, Gemini, and Claude, then reviewed sessions by landing page. That is the useful part of the example. The company did not treat “AI traffic” as one audience. It examined which source sent people to which page and what they did there. (refinelabs.com)

Another practitioner case makes the landing-page point directly. BrandSignal reported that ChatGPT became the second-largest referral source for one studio after the agency combined a citable knowledge hub with dedicated pages for high-intent services. The reported sample was 41 visitors and 48 page views, representing 14% of referred traffic. That is not enough to generalise. It is enough to show the mechanism: the knowledge content helped the business appear in AI-assisted research, while the service pages gave referred prospects somewhere more relevant than a homepage. (brandsignal.eu)

For a small business, that is the right way to read practitioner evidence. It is not a promise that AI traffic will suddenly fill your pipeline. It is evidence that a visitor who arrives with a specific expectation should not be sent to a page written for everyone.

A low-risk way to decide

1. Find the actual referrals

   In analytics, separate ChatGPT, Perplexity, Gemini, Claude, and other identifiable sources. Open the landing-page report instead of looking only at the channel total.

2. Read the destination page as a referred prospect

   Ask what the visitor likely already knows, what question remains unanswered, and whether the first screen confirms the reason they clicked.

3. Match the page to the buying situation

   Add specific proof, scope, pricing context, service area, process details, or comparison guidance when those details fit the referral intent.

4. Use a real conversion action

   Track calls, bookings, quote requests, purchases, or qualified enquiries. Time on page and scroll depth are supporting evidence, not the business outcome.

5. Keep the page useful outside AI

   The page should earn its place through customer usefulness, ordinary search visibility, internal links, sales enablement, or a measurable improvement in conversion.

What the page should contain

A good AI-referral landing page is usually less theatrical than a typical campaign page. It does not need a banner announcing that the visitor came from ChatGPT. It needs to confirm the decision context quickly and then remove the remaining doubts.

For a service business, that might mean opening with the exact job the firm handles, the type of customer it serves, the relevant geography, and the outcome it can reasonably support. Follow with evidence that fits the use case: a short example, named deliverables, a starting price or pricing logic, common constraints, and a clear next step. If the visitor is comparing providers, explain where your approach differs without turning the page into a vague superiority claim.

This is where your existing content can help. A local business that receives AI-assisted referrals about a buying question may need the kind of practical answer described in Local Businesses Should Answer Buying Questions Before Chasing Social Video Reach. A business comparing tools or approaches can use the structure discussed in Comparison Pages Are Gaining Value in Conversational Search, but Blog Posts Still Build Trust. The page does not have to be labelled an AI page. It has to make the visitor’s next decision easier.

Avoid building a page that merely repeats the same generic claims as your homepage. Google’s 2026 guidance specifically warns against “AI search” hacks, unnecessary content fragmentation, and inauthentic mentions. It says there is no ideal page length, no required special schema, and no benefit from rewriting content into an artificial format for AI systems. (developers.google.com)

The business case is conversion efficiency, not traffic vanity.

Suppose AI sends fewer visitors than Google. That alone tells you almost nothing. A smaller source can still deserve attention if the visitors are better matched to the offer, if the page converts their existing intent, or if the page also improves performance for referrals, sales follow-up, and ordinary search.

The reverse is also true. A high conversion rate from a very small number of AI visitors may be noise. It may reflect branded recommendations, repeat visitors, or a handful of unusually motivated prospects. Treat early results as a reason to inspect the journey, not as permission to move budget away from channels that already work.

The strongest investment is usually staged. First, make the relevant existing page clearer. Then create a dedicated page only when several referrals point to the same use case and the current page cannot serve it without becoming confusing. Keep the page crawlable, link it from the rest of the site, and give it a conversion path that does not demand a full sales conversation before the visitor has received a useful answer.

AI search changes how some prospects arrive. It does not change the basic job of a landing page: recognise the visitor’s situation, prove that the offer fits, and make the next action obvious. If a dedicated page does that better, build it. If it only gives a small traffic source a new label, improve the page you already have.