GEO Is a Distraction

An analysis of 12 months of Google Analytics 4 (GA4) data across 100 websites suggests that traffic from ChatGPT and other large language models (LLMs) remains limited compared with Google’s organic search traffic. The dataset covered more than 145 million sessions across websites of different sizes.

ChatGPT generated just 0.62% of the session volume driven by Google organic search. When traffic from all other LLMs was included, the figure rose to 0.73%.

Although monitoring whether AI tools mention or cite a website can provide useful insights, the data suggests that this traffic source remains relatively small for most businesses.

ChatGPT Visitors Convert Better but Google Drives More Sales

The analysis also examined conversion data from 55 of the 100 websites that tracked conversions in GA4. ChatGPT visitors converted at a rate of 13.8%, compared with 3.02% for visitors from Google organic search.

This means ChatGPT traffic converted at more than 4.5 times Google’s rate. However, Google organic search still accounted for 96.5% of the total conversions across the channels analysed. In this sample, Google generated 27 times more conversions than all LLMs combined.

These figures highlight the difference between conversion rate and overall business impact. While AI referrals may be valuable on an individual basis, their relatively low volume means Google remains the more significant source of conversions in this dataset.

The analysis also challenges earlier predictions that ChatGPT referrals would continue growing exponentially and eventually overtake Google. According to the data, growth in ChatGPT sessions across the 100 websites stopped accelerating in August 2025, around the release of GPT-5. Referral traffic subsequently declined.

Google Remains Central to AI Search

Generative engine optimisation (GEO), which focuses on improving visibility in AI-generated answers, has attracted considerable attention and investment. However, Marie Haynes argues that businesses may be overlooking a more significant opportunity by concentrating too heavily on mentions and citations in ChatGPT and similar tools.

AI is already changing how people search, but much of that activity takes place within Google itself. AI Overviews, AI Mode and AI-driven ranking systems allow people to use AI-assisted search without leaving Google’s platform.

Although this shift may reduce clicks to informational websites, Google remains an important route for people looking to find and engage with businesses.

Rather than focusing solely on rankings or AI mentions, Haynes recommends measuring meaningful business outcomes. Businesses should define what counts as a conversion or key event, track those actions and work towards improving them. In her sample, 45 of the 100 websites were not tracking any key events.

She also emphasises the importance of user satisfaction. Google’s ranking systems increasingly use AI to predict which results will be most helpful to searchers, making the quality and usefulness of a website’s content more important than simply optimising for machines.

The Bigger Opportunity Lies in Agentic Search

Google CEO Sundar Pichai has suggested that Search could evolve into an agent manager, coordinating tasks rather than simply presenting a list of links.

The future of search remains uncertain, but developments in AI agents suggest that websites and online businesses may need to prepare for new ways of being discovered, compared and accessed.

Google’s Sergey Brin has acknowledged that predicting what the web will look like in a decade is difficult. Demis Hassabis, meanwhile, has suggested that an agent-first web could operate quite differently from the pages people currently browse.

Several emerging technologies illustrate where these changes could lead.

Google Universal Cart

Google’s Universal Cart is designed to bring products from multiple retailers into a shared checkout experience within Google Wallet. It can monitor price changes, promotions and restocks, allowing shoppers to combine items from different sellers without visiting each retailer’s website.

The opportunity for businesses: Retailers should become familiar with Google Merchant Center and ensure their product information is accurate and up to date. Well-maintained product feeds can help Google’s shopping systems understand available products and potentially make them easier to discover and purchase through these experiences.

Universal Commerce Protocol

The Universal Commerce Protocol (UCP), backed by Google, Shopify and Stripe, is intended to help AI agents discover products, compare inventory and complete purchases through supported platforms.

Under this model, customers could buy products directly through Google’s services without necessarily visiting a retailer’s website, while the merchant remains the seller of record.

The opportunity for businesses: Technical SEO may need to expand to cover the ways AI agents discover and interact with products. This could include understanding UCP, its discovery endpoints and how product information and variations are represented through Merchant Center.

Product quality, competitive pricing and customer service will remain important. Businesses will also need to develop better ways to research competitors and understand their visibility as search results become increasingly personalised.

WebMCP

WebMCP, or Model Context Protocol for websites, aims to make interactive website functions available to AI agents. These could include forms, calculators, booking systems and other tools that allow agents to complete tasks on a user’s behalf.

The opportunity for businesses: Companies could explore which website functions would be useful for AI agents to operate. Examples might include a booking assistant, a pricing calculator or a self-service tool that helps customers find the right product or service.

Haynes believes it may be too early for many businesses to invest heavily in WebMCP-enabled tools, as the technology is still developing. However, experimenting with possible use cases now could help businesses prepare for future adoption.

Intelligent Eyewear

Google is also developing intelligent eyewear powered by Gemini and Android XR. These devices could introduce new ways for people to search for information and interact with businesses while going about their daily lives.

For example, someone could use smart glasses to identify a product they see, find it in their preferred size and purchase it. Other potential applications include guided cooking experiences or tools that help people troubleshoot household problems and connect with a professional when necessary.

The opportunity for businesses: Companies should consider how customers might discover or use their products and services in real-world situations. This could create opportunities for new shopping experiences, practical assistance and services that go beyond conventional website visits.

Although intelligent eyewear may not replace traditional search, it could expand the ways people interact with brands and make purchasing decisions.

Practical Steps for Businesses

Businesses do not need to invest in every emerging technology immediately. Instead, they can start by identifying where AI-driven search and commerce might affect their customers and testing practical ways to prepare.

Two useful starting points are improving the quality of product information in Merchant Center and exploring interactive website tools that could help customers complete tasks with less effort. These experiments can help businesses understand what is useful before making larger investments.

Teams should also establish clear conversion tracking so they can evaluate whether their search strategy is producing meaningful results. Rankings, AI citations and brand mentions may provide context, but they should not replace measures tied to business goals.

Focus on Outcomes Rather Than AI Mentions

AI is changing how people discover information and interact with businesses, but the data presented by Haynes suggests that chasing LLM visibility alone may not be the best use of resources for most websites.

Google continues to generate substantially more organic traffic and conversions than AI referral sources in the analysed sample. At the same time, Google’s own AI features and emerging agent technologies are creating new ways for people to discover products, compare options and complete transactions.

Rather than concentrating exclusively on GEO, businesses should strengthen their presence across Google’s search experiences, improve the usefulness of their websites and measure outcomes that matter. Exploring technologies such as UCP, WebMCP and intelligent eyewear could also help them prepare for an increasingly agent-driven web.

The next opportunity may not simply be getting an AI tool to mention a business. It could be making sure that AI-powered systems can discover its offerings, understand what it provides and help customers take action.