ChatGPT Ads: what the early data actually shows

ChatGPT Ads are now open to anyone. What was an invite-only beta — reportedly gated behind a steep monthly minimum — has opened up, and any advertiser can sign up, wait out an approval period of a week or two, and start buying placements inside ChatGPT’s answers. So the obvious question for any marketer is whether ChatGPT Ads are worth the money yet.

The honest answer, from the early data advertisers are sharing publicly, is: not for most of them — not yet. The platform is genuinely new, deliberately basic, and short on the things that make paid media accountable. That does not make it irrelevant. It makes it a surface to learn, carefully, with money you are prepared to treat as experimental.

This piece reads the early signals plainly, separates what is known from what is hype, and sets out how we think about advertising on AI assistants for clients.

What the platform looks like today

The interface is, by every early account, stripped back. You create a campaign, choose an objective of clicks or reach, pick a country from a short list, set a budget, and write a short ad: a title, a one-line description, a link and an image. That is most of it.

Two design choices stand out. There appears to be a bid floor of around $3 per click — bids at or above it are flagged as “strong delivery”, bids just below are warned they “may not deliver” — which reads less like optimisation guidance and more like a fixed minimum price for the network. And instead of keywords, you give the system “context hints”: a description of the conversations, topics or themes where your offer might be relevant. It is a hint to the model, not a targeting lever you control directly.

What the early numbers say

One advertiser’s publicly shared test — around $1,000 of spend on competitive B2B terms — is a useful, if small, data point. The reported figures: roughly $13 average cost per click, a click-through rate near 1%, an average CPM around $120, and zero conversions recorded over about 90 clicks.

A few caveats matter before anyone over-reads that. It is a single account, a single (expensive) B2B vertical, a tiny sample, and the advertiser noted their own UTM tracking was misconfigured early on, so even the click attribution was muddy. This is not proof the channel fails. It is an honest snapshot of how hard it is, right now, to know whether it works at all.

That uncertainty is the real story. The most consistent theme across early reports is not the cost. It is the absence of reporting. Advertisers describe seeing impressions, clicks, average CPC and CPM, and almost nothing else: no conversation themes, no audience breakdown, no path to optimisation. Conversion-based bidding is still marked “coming soon”.

The audience question no one can answer yet

Here is the structural issue that should give B2B advertisers pause. Ads only serve to free-tier users. Anyone on a paid ChatGPT subscription does not see them.

For a consumer brand, that may be a large and reachable audience. For B2B — where the senior buyers, founders and decision-makers you most want are disproportionately likely to be paying subscribers — it means the people you are trying to reach may be outside the ad inventory entirely. You are left advertising to whoever is on the free tier, and there is no published data yet on the quality of that audience for high-value B2B intent.

That does not kill the channel. It does mean the “right people at the right moment” promise is, for now, untested where it matters most for B2B.

How this fits the bigger picture: ads versus being the answer

It is worth stepping back. OpenAI’s leadership once framed advertising as a last-resort business model; the speed of this rollout, as the company moves toward a more commercial footing, tells you how central ad revenue is becoming. Paid placement inside AI answers is going to grow.

But paid placement is only one way to appear in an AI answer, and currently the less durable one. The other is to be the source the model cites organically, which is the discipline of generative engine optimisation and, practically, getting your company mentioned in ChatGPT without paying per click. One advertiser made the point bluntly: if the models already pull from and link to sources for free, optimising to be one of those sources may beat renting a slot at the bottom of the answer.

The mature view is that these are complementary. Paid AI placements buy presence now; organic AI visibility compounds and does not switch off when the budget does. A connected strategy uses both, weighted to what each can currently prove.

How Morris McLane executes this

AI advertising is a capability we run, not a trend we narrate. Our AI advertising capability places sponsored ads inside AI assistants like ChatGPT through the demand-side platform we operate, as one connected part of a paid-media strategy, never a standalone punt.

In practice, given how early the surface is, that means a disciplined approach to an undisciplined platform:

• Test budgets, framed as experiments. We treat ChatGPT Ads as a learning line with explicit, capped spend and zero-expectation framing, not a channel we push onto clients’ core budgets until it can prove return.
• Our own measurement layer over theirs. Because the platform’s reporting is thin, we instrument everything ourselves: rigorous UTM tagging, clean GA4 events, conversion tracking fired only on the right pages (not pasted site-wide, a common and costly setup error), and server-side data where it helps. The platform may not tell you what worked; our analytics will.
• Audience-fit judgement first. We assess whether the free-tier audience plausibly contains your buyers before spending. For some consumer and ecommerce briefs it may; for senior-B2B intent, we are honest that the audience question is unresolved.
• Paid placement alongside organic AI visibility. We run the ads as part of the same programme as your AI search visibility work, so you are building durable, cited presence at the same time as testing paid, not betting everything on a slot you rent by the click.

The throughline is accountability. The platform is too new to optimise in any meaningful sense; what we can control is how cleanly the spend is measured and how proportionate it stays.

The short version

ChatGPT Ads are open to everyone, but the early data — high click costs, a ~$3 bid floor, no conversion bidding, sparse reporting and a free-tier-only audience — makes this a test-budget channel, not a core one, especially for B2B. It will improve, and paid presence in AI answers will matter. For now, the smart play is to learn the surface with capped, well-measured spend while building the organic AI visibility that compounds. That balance, paid AI placement run with real measurement discipline, alongside earned AI visibility, is exactly what our AI advertising capability is built to deliver.