A US panel study surfaced a number that should recalibrate how teams think about ChatGPT as a commercial channel: just 6.9% of ChatGPT conversations showed any purchase intent. The majority of responses named no brand at all. If you have been building an AI search programme on the assumption that ChatGPT is a high-intent, bottom-funnel surface, this data puts a question mark on that premise.
That does not mean visibility in ChatGPT is worthless. It means you need to be precise about what visibility at different stages of the funnel is actually worth - and whether you are measuring it in a way that reflects that.
The Intent Mix Is Not What Most Brands Assumed
When ChatGPT crossed mainstream usage, a lot of brands rushed to frame it as a direct competitor to commercial search. The logic seemed sound: users ask questions, the model responds, brands that get named get considered. But that framing assumed the question mix looked something like Google Shopping or branded search. It does not.
The panel data points to a platform where the vast majority of queries are informational, conversational, or task-based - not purchase-driven. Users are drafting emails, summarising documents, exploring ideas. The 6.9% who show commercial signals are a minority, and even within that group, most responses name no specific brand. Getting cited at all is not the norm.
This matters because it changes where in the funnel your AI visibility programme should be doing its work. If the typical ChatGPT session is early-stage and exploratory, the value of a citation is likely closer to brand awareness than to last-click conversion. That is not a reason to ignore it - but it is a reason to stop attributing it like paid search.
Most AI Responses Name No Brand. That Is the Real Problem.
The finding that most ChatGPT responses mention no brand at all is arguably more significant than the 6.9% intent figure. It tells you that for the majority of queries - even those with some commercial dimension - ChatGPT is synthesising generic answers without reference to specific companies or products. Your brand is simply absent from the conversation.
This is a structural issue, not a content quality issue. AI models default to category-level descriptions when they lack strong entity signals about a specific brand. If your entity footprint is thin - limited structured data, weak third-party citations, no consistent factual presence across sources the model trusts - you are unlikely to be named even when the query is relevant to what you sell.
The implication for GEO practice is direct: earning a brand mention in AI responses requires that the model has sufficient, consistent, corroborated information about your brand to surface it confidently. Schema markup, authoritative backlinks, digital PR placements, and presence in the sources AI models are known to index - these are the inputs that move you from unnamed to cited.
What This Means for How You Prioritise AI Search Channels
Not all AI search surfaces carry the same intent profile. Perplexity, for instance, is used much more deliberately as a research and comparison tool. Google AI Overviews appear on queries that have already passed through the intent filter of a Google search. The query population arriving at each surface is meaningfully different, and that should shape where your optimisation effort goes.
If 6.9% of ChatGPT conversations show purchase intent, and your programme is investing heavily in ChatGPT visibility to the exclusion of Google AI Mode and Perplexity, the return on that effort is likely to be low and difficult to attribute. A more grounded approach is to segment your visibility programme by surface and map each surface to the funnel stage where it realistically operates.
ChatGPT citations earned on informational queries can still build brand familiarity. That is worth something - but it belongs in a brand awareness budget with brand awareness measurement, not in a revenue-attribution model that expects direct conversion signals.
Attribution Needs to Reflect the Journey, Not the Last Click
The measurement challenge this data creates is real. If a user encounters your brand in a ChatGPT response during an informational query, then returns a week later via a branded Google search and converts, the ChatGPT touchpoint is effectively invisible in most attribution models. GA4 will credit the branded search. Your CRM will show no AI channel influence.
This is not unique to ChatGPT - it is the same problem that existed with display and social before view-through attribution became standard. The fix requires the same logic: identify AI-referred sessions through UTM parameters where the platform passes them, segment chatgpt.com referral traffic in GA4, and look for patterns in time-to-conversion and query depth that suggest AI-assisted consideration journeys.
For lead generation businesses, the more useful signal is often in the CRM, not the analytics platform. If leads that convert at higher rates or higher values are more likely to have visited multiple informational pages prior to enquiry, that behavioural pattern may be an indicator of AI-assisted research - even if you cannot directly observe the ChatGPT session that started the journey.
The Practical Recalibration for AEO Teams
The 6.9% figure is not a reason to deprioritise ChatGPT entirely. It is a reason to be honest about what you are buying with AI visibility investment on that platform. Brand presence in a low-intent, no-brand-named AI response is worth something in an awareness context. It is not worth the same as a citation in a Perplexity comparison query or a Google AI Overview appearing on a high-intent search term.
For teams with limited resource, this data supports a channel-weighted approach: concentrate structured content, schema investment, and digital PR effort on the surfaces and query types where commercial intent is documented and measurable. Use ChatGPT visibility as a secondary signal rather than a primary revenue driver.
The brands that will get this right are the ones that treat AI search like any other channel - with clear funnel-stage assumptions, channel-specific KPIs, and attribution logic that matches how users actually move from an AI response to a purchase decision. That is a harder problem than optimising a page title, but it is the right problem to be working on.