OpenAI opened ChatGPT Ads to South African advertisers on 21 October with a minimum spend of $25 per day — roughly R410, or about R12,400 a month before bank fees. That low entry price is precisely why practitioners should treat the channel with discipline rather than urgency, argues Nawaaz Safodien, performance search account lead at Dentsu, in an analysis published 13 November.

Within days of launch, LinkedIn filled with agencies marketing themselves as ChatGPT Ads specialists, and clients began asking whether to shift budget. Safodien frames the real question differently: not whether the ads are worth testing, but whether consumers are moving from searching to asking — a shift that would extend far beyond one new media channel.

What actually launched?

Brands can now buy placements through OpenAI's self-serve Ads Manager. The ads appear separately from ChatGPT's responses, carry clear labels, and currently target users on the Free and Go plans. Financial services advertising is restricted to approved advertisers, reviewed case by case, and several policy analyses suggest access remains limited outside the United States. For banks, insurers and lenders, eligibility — not performance — is the first question.

What do the early numbers show?

Independent tests have produced mixed results:

• One retail advertiser recorded a 0.91% click-through rate, against 6.4% on Google Search in the same category.
• A side-by-side test found cheaper clicks on ChatGPT — $3.14 versus $6.76 on Google — but zero qualified leads, while Google produced two.

OpenAI's published partner case studies read more favourably. WeightWatchers reported a cost per acquisition 15.3% lower than paid search, and supplement brand Dose said 67% of incremental purchases came from new customers.

That last figure matters most, Safodien writes. If ChatGPT influences consumers during consideration — particularly people who have never engaged with a brand — then conventional performance metrics may miss the platform's actual value.

Why measurement is the trap

Safodien cites a dashboard he reviewed earlier this year that showed conversions up more than 100 times. The apparent breakthrough dissolved once his team established that most conversions were app logins rather than genuine acquisition events. A platform with no historical benchmarks is exactly where misleading success metrics thrive, he warns.

South Africa adds its own variables. Most published testing covers North America and Europe. Local consumer adoption of AI tools is growing quickly, but search behaviour remains entrenched, mobile dominates digital usage, and data costs still shape online activity.

How to structure a test

Safodien sets out six steps for practitioners:

1. Confirm eligibility. Verify that your category, business and market qualify, and that South African targeting is available.
2. Fix measurement first. Implement the pixel, configure server-side tracking, and define conversions before spending. A login is not a lead; a lead is not a customer; measure all three separately.
3. Ringfence a learning budget. Spend an amount you can afford to lose in exchange for insight, and treat phase one as research, not a growth channel.
4. Benchmark against your strongest search activity. Compare qualified leads, sales or customer acquisition cost — not clicks.
5. Define success before launch. Agree the thresholds that determine whether you scale, optimise or stop. Most tests fail because nobody agreed on success upfront.
6. Separate new and existing customers. If the early evidence holds, incremental customer acquisition could become the platform's strongest advantage; track it independently.

He adds one caution: avoid launching a first test during peak retail periods such as Black Friday, because noise makes learning difficult.

Should brands invest now?

"My advice to clients is not to be first, it is to be prepared," Safodien writes. "There is no prize for arriving first in a channel you cannot measure."

Whether ChatGPT Ads becomes a major media platform remains uncertain. What is clear, he argues, is that consumers increasingly ask AI platforms for recommendations, comparisons and advice they once sought through search. The real opportunity may not be buying ads inside AI ecosystems — it may be ensuring a brand stays visible, trusted and discoverable when those AI-driven conversations happen. That, he concludes, is the test to prepare for now.