Bianca Lourens is a Senior Marketing Account Manager at Distl, working with Australian businesses to align their digital strategy with business outcomes. Her work spans SEO, Google Ads, social, and web. Clients include Ancient Lakes Magnesium, The Chart & Map Shop, WA Good Food Guide, Fishbone Wines and Easystart Homes.

Content Overview

Current state of ChatGPT Ads
The cost of ChatGPT Ads has dropped a lot
What you can do with it right now in Australia
The results aren’t really there yet
The trust signal problem
Where I land

Current state of ChatGPT Ads

ChatGPT ads rolled out in Australia (alongside New Zealand and Canada) in mid-April, the first market outside the US to get them. Ads only show up for free-tier users and people on the new $13/month Go plan. Anyone paying for Plus, Pro, Business, Enterprise or Education sees nothing, which already narrows your reach.

The cost of ChatGPT Ads has dropped a lot

When it first launched in the US back in February, you were looking at a $60 USD CPM and a minimum spend commitment of around $200,000 to $250,000 USD. That priced almost everyone out.

It’s come down fast since then. The minimum spend dropped to $50,000 USD in early April, and then on May 5 OpenAI opened its self-serve Ads Manager with no minimum spend at all. CPMs are now sitting between $15 and $40 depending on category.

Cost-per-click bidding turned on with the self-serve Ads Manager launch in May, with the recommended range at $3 to $5 USD per click. Bid under $3 and you’re reportedly getting zero impressions. That puts ChatGPT roughly in line with Google Search and a bit above Meta on average, though it depends on the category. Software and finance run higher at $8 to $18, while ecommerce and retail sit at the lower end.

What you can do with it right now in Australia

(As of May 2026) Targeting is pretty light at this stage. Ads are matched to the topic of the active conversation, with personalised ads (on by default) factoring in past chat history. Geo-targeting is country-level only, so you can hit Australia but not Sydney or a specific postcode. OpenAI also blocks ads from running next to health, mental health or political topics, and anyone under 18 doesn’t see ads at all.

What’s coming later in 2026 is where it gets a bit more interesting: audience syncing and lookalikes (similar to the Facebook Custom Audiences setup), multi-turn conversation retargeting (reaching people who’ve had a real conversation about a problem you solve without converting), and CRM integrations with Salesforce, HubSpot and Microsoft Dynamics for closed-loop attribution.

The results aren’t really there yet

Click-through rates are sitting around 0.91%, roughly seven times lower than Google Search (around 6.4%). Similarweb has it even lower at 0.68% on average. The reason makes sense once you think about it: people on ChatGPT are in task-completion mode, not browsing mode. They’re likely trying to solve something and not be sold to.

The bigger issue is that early advertisers (the huge buyers like WPP, Omnicom and Dentsu) are reportedly struggling to prove their ads even worked. One ad buyer running an $800 million budget called the experience “worse than the early days of Google, Meta or even Snapchat.” A few big brands have already told their agencies they won’t come back without a trusted measurement partner in place.

The trust signal problem

This is the part I’d be most cautious about. ChatGPT users have built a real relationship with the platform as a helpful, conversational tool. When ads first appeared the backlash was pretty quick. OpenAI had to disable promotional app messages after users started sharing screenshots of ads that felt out of place. Their own chief research officer admitted they “fell short.”

That matters for your brand. If you advertise too early in a model people don’t yet trust as an ad platform, the association can feel forced. People need time to get used to the idea that ChatGPT is somewhere they get sold to.

Where I land

It’s a genuinely interesting landscape and I’m watching it closely for my clients. The good news is that the economics now make it accessible to almost anyone, and the targeting features rolling out later in the year could make it a lot more useful. Until measurement matures, user behaviour settles, and we start seeing proper case studies with real ROI, I’d rather keep your budget in channels we know perform.

This will likely be a real opportunity down the track. Just not right now.