You may have read that visitors who arrive from ChatGPT or Perplexity become customers far more often than visitors from Google. Some published figures point that way. They come from narrow samples, they measure different things, and none of them can tell you what will happen on your own website. Here is what those figures actually measured, and how to test the idea with your own enquiries.
What the published figures measured
Semrush published a study in July 2025 that describes the average visitor from an AI search tool as 4.4 times as valuable as the average visitor from ordinary search, based on conversion rate. The study does not say which websites were measured, how many, over what period or what counted as a conversion. It describes its traffic and value projections as extrapolations from historical data. Read the Semrush study.
Ahrefs reported in November 2025 that visitors from AI search made up about 0.5 percent of visits to its own website but about 12.1 percent of its sign ups. That is one software company counting free account sign ups on its own site, with no measurement period stated. The author says not everyone will get the same results, points to research that found ChatGPT referrals converted worse for online stores, and recommends running your own tests. Read the Ahrefs article.
These are useful signals. They are not a conversion rate you can expect. A sign up for a software tool, a purchase from an online store and an enquiry to a clinic, a showroom or a professional firm are different actions, made by different people, at very different prices.
Two explanations worth testing
There are reasonable ideas about why AI referrals might behave differently. Treat them as possibilities to test, not settled causes.
Buyer intent. Someone who asks an AI assistant to recommend a business for a specific need may be closer to a decision than someone typing a few words into a search box. They may equally be at the start of their research. The question they ask says more about where they are than the tool they use to ask it.
Trust in the recommendation. An AI answer often names a business and says something about it before linking to it. Some people give a named recommendation more weight than a list of links, and a visitor who still clicks may already know roughly what you offer. Others are rightly cautious about AI answers, and an outdated or wrong description can send people with the wrong impression.
How to test it on your own website
Decide what a conversion means for your business: an enquiry, a booking, a quote request or a sale. Count the same action for every source of visitors.
Check where visits come from. Many AI tools pass on the address of the page that sent the visitor, but some visits arrive with no source at all and are counted as direct traffic. Your count of AI referrals is therefore likely to be incomplete.
Ask new enquiries how they found you. A short question on the form or in the first conversation catches referrals that website reports cannot see.
Give it time. A handful of AI referrals in a month cannot establish a rate. Look at the quality of the enquiries as well as the number: did they become customers, and what was the work worth?
What this means for your website
Whatever brought them, a visitor who is ready to decide wants the same answers: what working with you involves, roughly what it costs, what you have done before and how to start. A website that answers those questions clearly serves people from Google, from an AI answer and from a friend’s recommendation.
On the figures above, AI tools still send a small share of visits; Ahrefs measured about half of one percent of its own. The foundations of good search work also help AI tools read and describe your business accurately. Treat AI search as a source worth measuring, not a reason to abandon what already works.
MM8’s services are built for businesses in many markets, from showrooms and clinics to professional services and trades. In each one, the useful question is not whether AI traffic converts better in general. It is whether it brings you good enquiries, and whether your website is ready when it does.
Sources
The Semrush and Ahrefs figures are described as their publishers report them, with the limits they state. The two explanations and the testing approach are MM8’s judgement. No conversion rate or result is promised.
This article was corrected in October 2026. An earlier version said AI search traffic converts about five times better than Google traffic and brings a higher average order value. Neither figure had a published source, sample or definition behind it, so both were removed.