Why ChatGPT Ads May Need a Different Measurement Framework

src-809c8a24

Marketers comparing advertising in ChatGPT with Google Search risk overlooking how conversational AI may influence buying decisions, argues Maor Sadra, CEO and co-founder of INCRMNTAL. In his view, judging the emerging channel mainly by clicks and cost per acquisition (CPA) applies search-era measures to a customer journey that works differently.

The distinction, Sadra says, is that search often responds to demand a consumer has already expressed, while a conversation with an AI assistant can help a person explore options and clarify what they want. That difference matters to advertisers evaluating whether conversational AI contributes value, particularly when a purchase happens after the conversation and through another channel.

Intent capture versus intent development

A search for “cheap hotels in Hawaii,” Sadra writes, signals a relatively clear commercial need. An advertiser can compete to be relevant when the person is considering a booking. A conversation might instead include details about the reason for travel, budget, accommodation preferences and activities. The user is not merely expressing a settled requirement; they may be working through priorities and comparing possible choices.

Under that model, an assistant could influence how a person defines the purchase before they visit a seller’s website. Sadra’s argument is that the advertising opportunity may therefore include participating in the development of intent, rather than only reaching consumers at the moment they demonstrate it through a query.

This is a conceptual distinction, not a guarantee that every conversation produces a purchase or that ChatGPT advertising will outperform search. The source of value, in Sadra’s framing, may be less visible to measurement systems that concentrate on the final click or the interaction nearest to conversion.

Why CPA may give an incomplete picture

Sadra acknowledges that early tests of conversational AI can show acquisition costs that look high relative to search. He says that was also his initial reaction. But he cautions against treating a higher CPA on its own as proof that the channel is ineffective.

A person who spends several minutes discussing products, asking follow-up questions and refining requirements may arrive at a retailer or brand site with a clearer idea of what to buy. That could make the eventual customer a stronger fit, Sadra argues, even if the cost of acquiring that customer appears higher. The point is not that such outcomes have been established across campaigns, but that CPA alone may not capture them.

This raises a familiar measurement tension: efficiency and effectiveness are related, but they are not interchangeable. A channel can look less efficient under one metric while potentially contributing to outcomes that the metric does not register. Advertisers need evidence about business results to determine whether that is happening in their own campaigns.

Attribution’s blind spot

For roughly 25 years, performance marketing has relied heavily on clicks, Sadra writes, in part because they are observable and can be used to optimize campaigns. Attribution models commonly assign credit to interactions close to a conversion, a practical approach when customer journeys are comparatively linear and interactions can be tracked.

Conversational AI complicates that picture when an influential interaction occurs well before a consumer clicks an advertiser’s link. If the person later converts through a different channel, traditional attribution may award credit elsewhere. That does not establish that the earlier conversation caused the purchase; it means, Sadra says, that a last-click view may fail to reflect any contribution it made.

He points to early evidence suggesting that measurement focused on incremental business outcomes can produce a different assessment of conversational AI than last-click attribution. He does not provide specific studies, figures or campaign results in the argument. For marketers, that makes the central recommendation a measurement question: test whether exposure to the channel changes business outcomes, rather than relying only on the conversion path captured by clicks.

A distinct channel, not a search substitute

Sadra frames the debate over OpenAI advertising as more than a contest over whether ChatGPT can match Google on CPA or conversion rates. Social, retail media and connected TV each prompted measurement adjustments, he notes, because their roles and consumer journeys did not map neatly onto earlier channels.

His case is that conversational AI deserves similar scrutiny on its own terms. Search advertising can monetize existing intent; conversational AI may have a role in shaping it. That difference does not make conventional metrics useless, but it does make them incomplete if used alone to judge the channel.

For advertisers and measurement providers, the practical challenge is to distinguish real incremental contribution from activity that merely precedes a conversion. Until that distinction is measured, Sadra argues, click-based comparisons may lead marketers to underestimate conversational AI—or to misread its performance in either direction.