.svg){kind=logo}
Updated 28 September 2026. Originally published 17 February 2026.
When we first wrote this post in February, ChatGPT ads were a US-only test with a handful of managed advertisers and a lot of speculation. Seven months later, ChatGPT ads run in more than 40 countries, including the UK, and OpenAI says the business passed a $1 billion annualised revenue run rate in under 200 days. We've also been running live campaigns for a UK SaaS client long enough to have real numbers.
So this is a rewrite rather than a refresh. Below you'll find what has actually shipped, how ChatGPT ads differ from traditional PPC, what our early cost data looks like, where the channel fits in a B2B funnel, and which companies should (and shouldn't) be testing it right now.
What's changed since ChatGPT ads launched
The first ads appeared on 9 February 2026 as sponsored cards at the bottom of answers, sold on a CPM basis through managed deals only. Since then the platform has filled out quickly.
OpenAI opened its self-serve Ads Manager in May with CPC bidding, a measurement pixel and a Conversions API, and no minimum spend. The UK became the first European market in June, with self-serve access following later that month. By the end of August, self-serve had opened across 31 European countries plus India, the Middle East and North Africa, and OpenAI reported "tens of thousands of advertisers" on the platform and more than 1 billion weekly ChatGPT users.
The September releases matter most for B2B advertisers:
- Conversion-optimised bidding is now generally available on both a per-click (oCPC) and per-impression (oCPM) basis
- Configurable attribution windows of 7, 14 or 30 days post-click, and 0 or 1 day post-view
- Custom audiences can now exceed 5 million members and mix identifier types, which makes CRM list uploads far more practical
- A HubSpot integration lets advertisers create ads, track performance and manage leads from inside HubSpot
- Sponsored Agents, a US-only test where clicking an ad opens a separate, labelled conversation with the advertiser's own AI assistant
What hasn't changed is the set of principles OpenAI published in January. Ads still sit below the answer and are clearly labelled. They don't influence what ChatGPT says. Advertisers still get aggregated reporting, not access to conversations. And ads are only shown to logged-in adults on the Free and Go tiers, so Plus, Pro, Business and Enterprise users never see them.
That last point deserves a moment's thought if you sell to enterprise. A buyer researching your category on a company ChatGPT Enterprise seat won't see your ad at all. We'll come back to why that makes organic AI visibility non-negotiable.
How ChatGPT ads differ from traditional PPC
On Google, a buyer compresses a problem into four or five words and you bid on those words. In ChatGPT, the same buyer describes the whole situation: their team size, the tool they're leaving, the budget they've been given, the integration they can't live without. The targeting unit isn't a keyword, it's the conversation.
In practice that changes how you build an account. There's no keyword list. Instead you write context hints, short descriptions of your product and the problems it solves, which OpenAI is explicit "are not exact-match keywords and do not guarantee delivery in specific conversations." You then layer custom audiences, geography and platform targeting on top, much as you would on any other channel.
The ad itself is also smaller than most search marketers are used to. A unit has an advertiser name, favicon, a short headline, a short description, an image and a landing page. There's very little room for clever copy, so the heavy lifting falls to relevance and to the page the ad sends people to.
And the auction is weighted towards relevance rather than pure bid. A tightly matched ad with a useful landing page can beat a bigger budget, which is part of why costs are currently so different from search.
What ChatGPT ads actually cost: our early numbers
We're running ChatGPT ads for a UK SaaS client in the office real estate sector, a category where paid search is brutally competitive. Over the months we've been live, average CPC on ChatGPT has come in at £1.96, against £36 on Google Ads for the same client. That's roughly 18 times cheaper per click.
Cheap clicks mean nothing if they don't turn into customers, so the number we care about more is acquisition cost. CAC through ChatGPT ads is £200, against £600 through Google Ads, a third of the cost for the same business.
Two caveats, because we'd rather you trusted the numbers than got excited by them. This is one client in one sector over a matter of months, so treat it as a strong signal rather than a benchmark. And OpenAI doesn't share conversation data, so some of the qualitative insight we'd normally pull from search terms reports simply isn't there yet. Pipeline-level tracking on your side is what keeps the channel accountable.
For context, OpenAI recommends starting CPC bids of $3 to $5. For a first B2B test we typically suggest £1,500 to £3,000 a month in media, which gives enough volume to reach a real conclusion in 8 to 12 weeks.
Where ChatGPT ads sit in the B2B funnel
It's tempting to label this a top-of-funnel awareness play. We don't think that's right, at least not for B2B.
By the time someone is asking ChatGPT to compare three HR platforms for a 200-person company with a German entity, they're well into consideration. They've defined the problem. They're building a shortlist. ChatGPT ads reach buyers in the middle of the funnel, at the moment the shortlist is being formed, which is often before they've opened a search tab at all.
That's also why our CAC numbers look the way they do. The conversation has already done a lot of the qualification work that a broad search keyword can't.
Using ChatGPT ads for lead generation
For lead gen, the practical setup looks a lot like any well-run B2B account. Install the OpenAI pixel and Conversions API, define the conversion events that matter (demo requests and qualified sign-ups, not page views), and use oCPC or oCPM once you have enough conversion volume for the model to learn from. Longer attribution windows help here, since B2B buyers rarely convert on the first visit.
The step most teams skip is closing the loop with the CRM. If your ChatGPT leads land in HubSpot or Salesforce, track them through to opportunity and closed-won, and judge the channel on pipeline rather than platform-reported conversions. We apply the same offline conversion approach across every channel we run, and ChatGPT ads is no exception.
Landing pages matter more than usual too. Someone arriving from a detailed conversation expects the page to pick up where that conversation left off: plain language, pricing or a clear route to it, proof above the fold, and an obvious next step. A generic brochure page wastes the context the buyer arrived with.
Which industry verticals suit ChatGPT ads best
From what we're seeing, ChatGPT ads work best where three conditions overlap: buyers do heavy research before talking to sales, the decision involves several stakeholders, and Google CPCs are painfully high.
That points to a fairly clear set of B2B verticals:
- Proptech and workspace software, where our client data comes from and search CPCs regularly run past £30
- Fintech and payments, where buyers compare fees, compliance and integrations in detail
- HR, payroll and people software, where questions are highly situational (headcount, countries, existing stack)
- Cybersecurity, one of the most expensive search categories there is, with technical buyers who already use AI tools heavily for research
- Martech and developer tools, whose buyers are some of ChatGPT's heaviest users
Is it right for Series A+ companies and scaleups?
ChatGPT ads are a strong test for Series A and later companies that already have paid search working. They're not the right first channel.
The companies that get the most out of it tend to share a few traits. They already run Google or Microsoft Ads with reliable conversion tracking. Their CRM is connected, so they can judge a new channel on pipeline. They have a defined ICP and enough budget to ring-fence a test without starving channels that already deliver. And, often, they've hit the point where search CPCs are climbing faster than conversion rates, so a cheaper route to the same buyers is worth real money.
Scaleups expanding internationally have an extra reason to look now. Self-serve access opened across 31 European markets in August, and the same account can target by country, which makes ChatGPT ads a relatively low-cost way to test demand in a new market before committing to full search builds.
If you're pre-Series A with no paid media history, the better move is still getting Google foundations right first. You'll learn more, faster, from a channel with mature reporting.
Why ChatGPT ads shouldn't run in isolation
Here's the thing most early coverage misses. ChatGPT ads sit below the answer. The answer itself, the recommendations, the shortlist, the "here are five tools worth considering", can't be bought. It's earned through organic AI visibility.
If a buyer asks ChatGPT for the best workspace management platforms and your competitor is named in the answer while you appear only in a sponsored card underneath, you're paying to be the alternative. Flip that around and the ad reinforces a recommendation the buyer has just read, which is where it performs best.
Remember the enterprise point from earlier as well. Business and Enterprise users never see ads, so for many enterprise buyers, organic visibility is the only way you'll appear in ChatGPT at all.
That's why we plan ChatGPT ads alongside organic AI visibility work, not instead of it. In practice that means clear, specific content that AI models can cite, consistent facts about your product across your site and third-party sources, structured data, and presence on the review and comparison sites models lean on. Paid gives you control and speed. Organic gives you credibility inside the answer. Each makes the other work harder.
The same logic applies across your paid mix. ChatGPT ads work best planned next to Google, Microsoft and LinkedIn, with budget moved according to what the CRM says each channel actually contributes.
What's coming next for ChatGPT ads
A few developments are already signposted. OpenAI has said it plans a conversion optimisation model that accounts for both click-through and view-through conversions, which should suit longer B2B journeys. Sponsored Agents are being tested in the US and, if they roll out further, will turn the ad click into a conversation with your own AI assistant, a genuinely new lead capture format for B2B. Opt-in features that adapt headlines to the conversation are also arriving.
Our view on costs is simpler. Today's CPCs reflect an early, uncrowded auction, and that won't last. Tens of thousands of advertisers have joined in seven months, self-serve is open across dozens of countries, and integrations with HubSpot and Shopify are lowering the barrier further. We'd expect CPCs to climb as competition builds, which is exactly why the teams testing now will have the learning, the audience data and the conversion history when prices rise.
If you're already running paid search and want to know whether ChatGPT ads belong in your mix, and how to make sure you're also showing up in the answers above them, get a free proposal and we'll look at where it fits for your account.
