Mathew Schwartz via Unsplash
Australian independent agencies are backing ChatGPT's new advertising platform as a strategic priority, warning that waiting for measurement tools to mature will already be too late.
OpenAI's pilot ad program launched in Australia on April 17 by invitation, with CPMs ranging from US$15 to US$60, no ROI measurement tools, no prediction models and no performance guarantees.
Independent agencies were among the first invited.
Robert Tadros, CEO and founder of Impressive, said the absence of measurement was beside the point.
"We're not pretending it's a performance channel, it's not. We position it as a strategic land grab. If you're waiting for perfect attribution, you're already too late," said Tadros.
"We're tracking directional impact: branded search, direct traffic, assisted conversions. If demand is moving, it's working. Simple as that."
OpenAI reduced its minimum investment from US$200,000 to US$50,000 since its US launch, yet Tadros said the lower price point was not what drove the decision.
"If a business is doing it because it's cheap, you've already missed the point. The real value is being early in a channel that's going to reshape how people search and make decisions,” said Tadros.
“This is early Google all over again - the upside is in being first, not being efficient.”
Angela Hampton, Founder and CEO at WiredCo., said the decision was about securing a seat at the table before the channel matures.
"Early bird gets the worm. For us and our client partners, participating in the trial was about getting a seat at the table early on," said Hampton.
"It gives us the opportunity to not only help shape how AI-native ads evolve in Australia and work directly with OpenAI, but also to get in early so we can start developing our strategies and experience in what will most likely become a huge advertising channel in Australia.
"In terms of metrics for success, we have been very real with our clients about there being no guarantees or prediction models we can use yet to estimate performance this early on. For us, it's less about performance at this stage, and more about having the opportunity to learn, evolve, and build a direct relationship with OpenAI, who so far have been legends to work with," said Hampton.
OpenAI's own pace of development is part of the attraction.
Nick Hayes, head of digital at The Media Store, said the platform's speed of iteration was unlike anything he had seen at this stage of a new ad product.
"It amazes me that, within eight weeks of going live with ads globally, OpenAI have a self-serve ads manager, and that we're onboarded and have the ability to trade. When discussing forthcoming changes to the platform, the ChatGPT team noted they were working on a conversions API product that they hoped to release in the 24 hours following the call," said Hayes.
"Since onboarding I've received multiple updates to their privacy policy. They've made it clear they want to learn and iterate on the fly and I think that's the draw to indies who can be nimble like this."
The cost picture is more complicated.
Hayes said the CPMs were high relative to comparable placements.
"The minimum investment levels are very accessible for indies and clients who want to test something new without the heavy price tag," said Hayes.
With that said, the CPMs are quite high - US$15 to US$60 - relative to the ad unit itself and our current visibility of the impact it has on the consumer journey. The ad unit is very reminiscent of other high impression, low CTR in-feed and responsive placements that trade at a fraction of the price."
The format itself remains limited. Hampton said the current unit was so basic her team had stopped calling the image an image altogether.
"The trick, apart from nailing the campaign structure, is to write the copy as natively as possible and have strong, highly relevant landing pages set up to send the traffic to. We're also playing around with reactive culturally relevant copy too," said Hampton.
Hayes raised a deeper question about what targeting means in an AI-native environment.
"We were told the main thing impacting performance is the ad copy. This begs the question - what's the role of keyword targeting in platforms like ChatGPT and paid search when long tail is moving away from a sentence and more towards a paragraph," said Hayes.
"Is semantic recognition and training your ad algorithm like a model more important than keywords in the future search-adjacent media landscape? Advertisers have been getting used to the concept of keyword-less campaigns through products like PMAX for a while now. It feels like we're closer than ever to ripping off that bandaid."
Tadros said the speed at which independent agencies had moved illustrated a structural advantage over holding companies.
"This is exactly where independents win. No playbook, no benchmarks - that's where big networks stall. Indies move. We don't need perfect data to make a call. We test, we learn, we double down. That's the edge," said Tadros.
For agencies yet to receive an invitation, Hayes offered a note of caution.
"From a risk perspective, I'd say read the terms and conditions - they are structured to give OpenAI autonomy to move quickly and change things. I'd also drip feed budget and scale with measurability once measurement matures," said Hayes.
"Those with marketing mix modelling in place or other predictive measurement tools are already best set up to test ChatGPT's impact on their campaigns."
Have something to say on this? Share your views in the comments section below. Or if you have a news story or tip-off, drop us a line at adnews@yaffa.com.au
Sign up to the AdNews newsletter, like us on Facebook or follow us on Twitter for breaking stories and campaigns throughout the day.