OpenAI introduced cost-per-click bidding for ChatGPT ads in April 2026. Early reports suggest CPC rates between $3 and $5, though these figures remain unconfirmed and should be treated as indicative rather than definitive.
The shift matters. It signals where OpenAI wants to take the platform.
For most NZ marketing managers, it changes nothing about what they should be doing right now.
What We Already Knew
ChatGPT ads rolled out to New Zealand and Australian users earlier this year. The US beta programme that preceded it required a minimum spend of $200,000 USD with CPM rates around $60.
There was no conversion tracking. No revenue attribution. No pathway to connect an ad impression to a downstream business outcome.
The barriers were high. The measurement was blunt. For most NZ businesses, the platform existed in theory but not in practice.
What Just Changed
OpenAI has now introduced CPC bidding alongside the existing CPM model. The CPM rate appears to have dropped significantly from the US beta level, though precise figures for New Zealand markets remain unconfirmed.
OpenAI is also actively hiring for advertising measurement infrastructure. The company is building attribution capabilities from relatively early in the platform's commercial life.
That timing matters. Uber didn't hire its first head of measurement until approximately three years after introducing its ad business. OpenAI is moving faster.
Why the CPC Shift Signals Intent
CPM is a reach model. You pay for eyeballs. Whether those eyeballs did anything is someone else's problem.
CPC is fundamentally different. You only pay when someone clicks. That pricing structure requires confidence in the value of the click itself.
Anyone who bought Google Ads in the early days will recognise this pattern. Platforms start with blunt instruments because that's what they can sell before the measurement infrastructure exists. As the platform matures and advertiser confidence grows, they move toward performance-based models.
The CPC introduction tells you OpenAI is serious about becoming a performance channel. They want a seat at the media planning table alongside Google and Meta.
The question for NZ marketing managers isn't whether that ambition is credible. It probably is. The question is whether the platform is ready for their money right now.
The Measurement Gap That Still Exists
ChatGPT ads currently report impressions and clicks. That's it.
There's no conversion tracking, no revenue attribution, no way to connect a click on a ChatGPT ad to a sale, a lead, or any downstream business outcome.
CPC tells you what you paid per click. It doesn't tell you what those clicks were worth.
In Google Ads, CPC only became useful once conversion tracking existed. Before that, you knew your cost per click but had no idea whether you were buying good clicks or expensive waste.
ChatGPT ads are at that pre-conversion-tracking stage right now. The pricing model has matured faster than the measurement has.
For a large global brand running awareness campaigns, that's manageable. For most NZ marketing managers working with accountable budgets, it's a genuine blocker.
You can't run a performance campaign on a platform that can't tell you whether it performed.
The Volume Reality
AI platforms are driving an average of just 1% of overall web traffic across ten major industries.
That means the total addressable audience using ChatGPT for search behaviour that might convert is still small relative to Google.
People still use Google for transactional searches. The behaviour hasn't shifted yet at scale.
Before committing budget to a new channel, a marketing manager needs confidence that their actual customers are using the platform with enough frequency and in the right context to make it viable.
That varies enormously by category. Some audiences are heavy ChatGPT users. Others barely touch it.
Right now most businesses haven't done that analysis. The platform doesn't provide the audience insight tools to make it easy.
Why GEO Remains the Priority
Organic AI visibility builds something you own. Paid access rents you visibility for as long as you're paying. The moment you stop, you're gone.
When ChatGPT recommends your business organically, it carries the weight of the platform's credibility. The user asked a question and the AI decided you were the right answer.
An ad appearing alongside that answer feels different. One feels like a recommendation. The other feels like an interruption.
GEO works across all AI platforms simultaneously. The entity presence you build to appear in ChatGPT organically also improves your visibility in Perplexity, Google's AI Overviews, and whatever comes next.
Paid access on ChatGPT only buys you presence on ChatGPT. The underlying work doesn't transfer.
GEO takes time to build. Entity associations, earned media signals, cross-web consistency, none of that happens quickly.
Businesses that start now are building a foundation that will be significantly harder to replicate in two or three years when the category is more competitive.
ChatGPT ads will still be available then. The organic visibility window is open now but it won't stay this accessible forever.
If you haven't already, check what AI platforms currently say about your business. Run an entity audit to see where you appear and what gets recommended when someone asks about your category. If you'd like a hand with that, ADMATIC offers a free AI search visibility audit for Marketing Association members.
What Would Change the Recommendation
Three signals need to exist before the advice shifts from "watch and learn" to "test seriously."
Conversion tracking that connects a click to a downstream business outcome. Not just clicks and impressions but an actual pathway from ad interaction to lead, sale, or booking. That's the non-negotiable.
Confirmed NZ pricing and access terms that make a meaningful test possible without a six-figure minimum commitment. When a mid-market NZ business can scope a genuine test within a realistic experimental budget, the conversation changes.
Independent evidence on format performance. Not OpenAI's own case studies, but documented results from advertisers who've run campaigns, reported outcomes, and had those outcomes scrutinised.
When those three things exist simultaneously, the recommendation changes. Not "go all in" because that's never the right call on a platform still finding its feet. But "build this into your media planning conversation" rather than "watch from the sideline."
The timeline will depend on how quickly OpenAI builds out the measurement infrastructure and confirms local market terms.
That's the framework.