Advertising has long followed a familiar sequence of catching attention, earning the click, driving consideration and eventually, converting interest into a purchase. Now, OpenAI’s Sponsored Agents could complicate that sequence by putting the brand inside the conversation itself.
On September 16, 2026, OpenAI launched Sponsored Agents, allowing people to start a conversation with a business-sponsored agent after clicking on an ad on ChatGPT. The company has also introduced new AI creative tools in Ads Manager and announced integrations with HubSpot, its first CRM partner, and Shopify, its first e-commerce partner, bringing ChatGPT advertising closer to the tools businesses already use. The format is currently being tested with select advertisers in the US. For the advertising industry, however, the bigger question is not simply whether another ad format has arrived. It is whether the ad itself is beginning to behave differently.
When awareness, consideration and conversion happen together
Karthik Shankar P S, Independent Brand & Growth Strategy Leader, sees the idea as genuinely interesting, but believes the industry needs to separate the promise of the format from the maturity of the infrastructure supporting it. “OpenAI's Sponsored Agents are genuinely exciting. Being excited doesn't mean being ready,” he says. He explains the proposition through a simple example. A consumer looking for a dining table could encounter a Wayfair agent that answers questions about dimensions, finish and delivery just like a knowledgeable salesperson walking up to them at the right moment. But, he cautions, “interesting ideas and mature infrastructure are two separate conversations.”
Amyn Ghadiali, Country Head, Gozoop Creative, believes the format could put the traditional marketing funnel under pressure. His trust has long been in ‘human-led, AI-enabled’, and he sees Sponsored Agents as one of the first formats to actually test that proposition. “India's the proving ground: 100M weekly users, ads live since 27 August, 50+ brands in week one,” Ghadiali says. He believes Indian creativity has moved from simply chasing fame towards effectiveness driven by cultural relevance. A sponsored agent makes that effectiveness more tangible because there is no impression to admire. There is an answer that has to survive follow-up questions.
“Awareness, consideration and conversion stop being sequential stages and start happening in the same breath,” he adds. That could put even established brands under a different kind of scrutiny. Brands with years of recall may now have to prove that the equity they have built can hold up when a consumer starts questioning, comparing and probing the product in real time.
Abhirup Datta, CEO, Performance Practice Media Solutions, dentsu India and CEO, Sokrati India, believes this is where the format could have its biggest impact. “This is the real shift as influence moves closer to the point of purchase,” he says. Today, a consumer may see an ad, conduct research separately, compare products and eventually make a decision. With sponsored agents, the brand can potentially enter the conversation when that consumer is already asking questions or comparing options.
From competing for attention to competing for trust
That changes the nature of what the brand needs to deliver. Harish Bijoor, Business & Brand Strategy Specialist and Founder, Harish Bijoor Consults Inc, sees an even broader change coming as AI agents enter marketing, branding and advertising interactions. “There is a huge change waiting ahead for us with the AI agents entering into the space of marketing, branding, advertising, and advertising interaction,” he says. According to him, AI could change how consumers are acquired, how attention is captured, how interest is developed and how desire to buy is created. But he also flags a potential downside: competing brands could deploy competing bots, creating an environment where the consumer is effectively navigating one AI agent against another. The eventual question, he says, could become: “Which bot will you believe?”
Vikas Chawla, Co-Founder of Social Beat, takes a more measured view of how much the traditional advertising funnel is actually changing. He believes the funnel doesn't change but the middle gets compressed. The ad still needs to earn the click, but the agent comes after it. What could change is the messy middle where consumers research, compare and shortlist products across multiple sources. “Shortlisting now happens inside one conversation and the brand does not have full control over this,” he adds.
This also puts greater emphasis on Generative Engine Optimisation, or GEO, as brands attempt to ensure their products and content are surfaced appropriately within AI-led discovery. For agencies, that means creative cannot simply mean producing another set of films or static assets. Chawla thinks brands will need a broader content system that caters to different audience personas, while media planning becomes more complicated as parts of the consumer journey move inside LLMs and remain opaque.
Shankar sees another creative challenge emerging: the brand’s conversational voice. He asks: what does your brand sound like mid-conversation, when someone is thinking out loud and not yet decided? While a headline makes a statement, a sponsored agent has to sustain a dialogue. He points out that brands' conversational voices are going to feel that gap.
The metric problem
Perhaps the biggest challenge is not creative but measurement. Sponsored Agents may still generate familiar performance metrics such as click-through rates, conversions and ROAS. But experts question whether those numbers capture what is actually happening inside an AI conversation.
Datta argues that CTR and ROAS could become less straightforward as measures of influence. “CTR measures attention on a static unit. But if influence is now happening inside a multi-turn conversation, closer to purchase, the click in this case tells you nothing about whether the conversation moved someone toward buying,” he says. ROAS presents a similar problem. In a multi-turn interaction, there may be several points at which influence occurs, making it harder to identify which interaction actually contributed to the eventual purchase.
He believes agencies will need to look at metrics such as lead quality, conversation quality and holdout testing against consumers who did not see the agent. “This is still early enough that nobody's agreed on what a good outcome looks like, so we shouldn't over-trust the reported numbers yet,” he says.
Ghadiali believes the bigger change is likely to be a move away from simply measuring efficiency towards understanding the density of the signals generated during a conversation. “A conversation with zero clicks can carry more purchase signals than ten banner clicks ever did. CTR is measuring the wrong axis,” he says.
He argues that brands have already been pushed to look beyond vanity metrics and focus on measures such as mindspace and brand salience. Conversational advertising, he believes, raises the stakes because even the basic economics of the format are still evolving. “With India's users overwhelmingly free-tier, whoever writes the ‘signal density’ equivalent for conversational ads owns the next measurement standard and it will likely be established here first,” Ghadiali adds.
Chawla expects CTR to have lesser relevance in LLM environments, with more attention shifting towards what happens within the conversation: the questions asked, the qualifiers consumers want to check and the information they seek before making a decision. The problem is that platforms are not yet providing that level of visibility. “There is no brand conversation-level attribution yet,” Chawla says, describing the current environment as a black box, similar in some ways to Meta Advantage+ and Google Performance Max.
Shankar makes a similar distinction between having numbers and understanding them. “The numbers exist. CTR, ROAS, conversions. OpenAI's Ads Manager reports all of it. The problem is what sits underneath them,” he says. On traditional search platforms, advertisers can understand the search term that led to an ad click and use that information to refine future campaigns. With ChatGPT Ads, the advertiser does not see the conversation surrounding the ad. “You know someone clicked. You don't know what they were asking, what the AI said around your ad, or what mindset they were in,” Shankar says. That makes optimisation harder because the data may show what happened without explaining why.
He also points to the limited public evidence available so far. The only published real-spend data, according to Shankar, comes from a 15-day campaign that delivered around 1.5x ROAS, from campaigns that were already performing. OpenAI has not yet released cross-advertiser benchmarks. “You can spend money here. Optimising that spend with any confidence is a different matter,” he says.
The agency model gets rewritten
The implications for agencies therefore extend beyond buying another media format. Datta believes planning will move closer to the actual purchase decision, with greater emphasis on understanding the specific needs and situations consumers are trying to solve at the moment of comparison. Creative will have to become more useful, rather than simply attention-grabbing. And if a brand's agent is handling the conversation independently, agencies will also have to anticipate how it responds to objections, comparisons and difficult questions.
“The clearest pattern so far: structure and message discipline matter more here than budget size, because you're being judged in real time, right at the decision point,” Datta says.
Bijoor expects the role of AI agents to eventually extend even further, with bots potentially involved in media planning, audience targeting, attention, interest, desire and eventually sales. That could create a new competitive environment where brands are not just competing for consumer attention, but where their AI agents are effectively competing for consumer trust.
For now, however, the US pilot remains the critical test. The industry will be watching what Wayfair and other early advertisers learn, whether consumers engage with branded agents after the novelty wears off, how much trust they place in sponsored answers and whether the format can demonstrate measurable commercial value.
Sponsored Agents may bring advertising closer to the point of purchase. But that does not automatically make the format a replacement for everything that came before it. As Shankar puts it, the development is worth watching closely. It may represent genuinely new creative territory, but it is too early to reorganise a media mix around it.
The ad may have started talking back. The industry is still figuring out what happens when consumers answer.
