ChatGPT began running ads in Australia and New Zealand in late March 2026, marking the first time a major LLM has sold advertising at scale. Japan and South Korea are next, with access coming "in the coming weeks." No timelines have been given for Hong Kong, Singapore or other APAC markets. OpenAI has said it plans to expand the format globally through 2026.
OpenAI’s head of global ad solutions, Dave Duggan, said the expansion follows “strong interest from businesses looking to reach users in a more conversational, intent-driven environment.” He added that as access broadens, OpenAI remains committed to “answer independence, privacy, and user control.”
The central question for brands at this stage is what advertising look like inside an answer engine?
"We are already starting to get access to these products in Australia and the early signs are interesting," says Shai Luft, co-founder at Bench Media. Australian agencies like PHD echo this in early tests.
"Early tests show highly engaged users, but creative must feel genuinely helpful rather than promotional," says Gemma Dawkins, national head of Digital at PHD Australia. "The format rewards brands that contribute to the conversation."
According to Luft, the kind of engaged traffic that answer engines offer is what agencies are banking on, because users are already in a consideration mindset rather than passive browsing mode. But he also warns that the space comes with major measurement gaps. “Reporting is currently very limited, with most environments only offering impressions and click data. Conversion tracking, attribution and quality-of-traffic measurement are still underdeveloped, which means marketers need to get comfortable operating with incomplete data while the ecosystem catches up.”
That is why some in the market are urging brands not to wait for a fully mature product before preparing. Lars Maehler, client lead of Publicis Media Hong Kong, says the smartest move is not to wait for every market to flip the switch, but to build the foundation that makes any of these placements work, and work better than they would in isolation. “The brands that will perform strongly in paid LLM environments are the ones that already have strong, consistent signals of Experience, Expertise, Authoritativeness and Trustworthiness (E-E-A-T) across multiple sources,” he said. “That means tightening content strategy, structured data, expert-authored pieces, and genuine review/sentiment management, not as ‘SEO or GEO hygiene’ but as crucial initiatives to drive overall visibility in an agentic web.”
Maehler recommends running small, well-scoped pilots where inventory exists, but treating them as learning exercises in a new intent layer rather than scale plays. “The agencies that moved early in ANZ are already treating this as a distinct creative and measurement challenge, not ‘Google Ads with better context’.”
Others are framing the shift even more broadly. “Marketers should stop thinking about LLM advertising as a new channel. It’s a clear signal that the search model itself has already changed,” said Yi En Chye, VP, experience & activation, APAC, Assembly. “We’re moving from a world of keywords and SERPs to one where AI engines deliver answers, not options. The brands that win will not be the ones who move fastest on LLM ad spend. They will be the ones who show up consistently in the answers that matter.”
Creative, targeting and measurement
From an operational standpoint, the critical first step involves data hygiene. “Brands should prioritise ensuring strong data hygiene across websites, product feeds and socials, which will make their content crawlable and their products surface within the answer engines,” says Sam Werngren, senior business director at VaynerMedia APAC.
For organic strategy, brands must leverage tools tracking engine visibility and high-potential prompts to adapt content. Product feeds should integrate directly into ChatGPT’s merchant centre, while for Gemini, YouTube is an increasingly important source of information, making titles, descriptions, chapters, timestamps and transcripts essential.
Meanwhile, paid strategy requires a completely separate mindset. “Brands need to set aside time to understand how answer engine targeting works, as it is different from traditional search,” says Werngren. “Forward-thinking brands who make such decisions now will hit the ground running as soon as paid opportunities become available in their region.”
Shifting to the strategic timeline, Laura Prieto, senior digital director at UM Australia, views this early phase as an essential learning period. “We are still in the earliest phase of LLM-based advertising, so the most important thing marketers can do now is test early and systematise those learnings. The advantage right now isn’t scale, it’s learning velocity.”
Creative execution and targeting must become far more adaptive. “Creative will have a significant nuance as brands need to approach LLM platforms differently to traditional search ads, this isn’t a lift-and-shift from traditional search,” adds Prieto. “Queries are more complex, nuanced and often deeply personal, so creative needs to be more adaptive, assistive and context aware.”
Opportunities and risks
Conversational AI allows brands to show up in highly engaged moments in the user decision journey, from shaping initial consideration to reinforcing loyalty when users are weighing alternatives. But that same dynamic also creates the biggest risk. If brands fail to show up in a way that is genuinely useful and contextually aligned, they simply won’t appear, or worse, they will feel intrusive and erode trust.
Above all, LLM advertising is uniquely powerful because these platforms sit incredibly close to final decision-making. “If an AI assistant recommends your brand while helping someone solve a problem, that is potentially more influential than a traditional search ad or social impression because it carries an implied layer of trust and recommendation,” says Luft.
At the same time, Luft warns against overreliance on early placements. “We risk replacing the open web with a handful of AI gatekeepers quietly deciding which brands deserve visibility. And unlike search engines, users may never even see the competing alternatives.”
Formats that fit
The formats most likely to work are the ones that feel native to assistance and recommendation. Winning formats will actively add value to the answer rather than disrupt it, pointing to contextual product suggestions as the most promising starting point.
“Brands should maximise all available real estate within available ad formats by integrating product feeds so that images, descriptions and pricing are available to consumers,” says Werngren. “That being said, I still expect paid answer engine responses to be a worthwhile investment.”
Formats like sponsored responses, contextual suggestions, and AI-guided comparisons make sense because they align with existing user behaviours. Longer term, the grander commercial opportunity isn't traditional advertising at all, but AI agents influencing purchases and brand decisions directly on behalf of users, a shift that could be commercially powerful but ethically murky.
The ultimate challenge for platforms, Luft predicts, will be balancing monetisation with trust. “The second users feel recommendations are being manipulated too aggressively by advertisers, the credibility of the entire product starts to break down.”
Maehler adds that the brands set to win aren’t the ones with the biggest budgets on day one, they’re the ones that have already invested in coherent brand ecosystems, strong trust signals across sources, and content that answers real questions clearly and authoritatively.
"Paid placements in these environments then become powerful amplifiers rather than the foundation," says Maehler. "Start building that foundation today. Watch the ANZ learnings closely. And approach this with the same curiosity we’re applying to AI more broadly: not just 'how do we do what we already do, but cheaper and faster?' but 'what new value or experiences can this unlock?'”
