What ChatGPT Ads means for B2B marketers, and why it doesn't reach them yet
ChatGPT now processes roughly 2.5 billion prompts a day and reaches around 900 million weekly users. Separately, G2's 2026 survey of over 1,000 B2B software buyers found 51% now begin their research with an AI chatbot more often than with Google, up from 29% in April 2025. Put those two facts together and ChatGPT Ads look like an obvious channel for B2B marketers to move into.
Then one B2B paid media agency actually ran the test. A $33 daily budget, nine ad groups, around 100 topic-based targeting signals scoped tightly to their category. After ten days, the campaign had spent $34.89 total and generated ten clicks. Not because the ads underperformed, but because there weren't enough relevant conversations happening on the platform to spend the budget on in the first place.
That gap between the headline numbers and the lived experience of actually running a campaign is the real story here.
The reach problem is structural, not temporary
OpenAI's own documentation confirms ads currently serve only to users on the Free and Go ($8/month) tiers. Anyone on Plus, Pro, Team, Business, Enterprise, or Edu plans sees no ads, no matter how well a campaign is targeted.
ChatGPT LLM
OpenAI are the company behind ChatGPT
Globally, that's not a small population, an estimated 81% of ChatGPT's total user base sits on the free tier. But B2B tech and SaaS buyers don't look like the general population.
One agency's own LinkedIn poll of 100 SaaS and marketing professionals found 74% were already on a paid ChatGPT plan, with only 26% on free, nearly inverting the platform-wide split. If your buyers are marketing leaders, demand gen directors or senior operators, there's a real chance most of them have already paid to opt out of seeing your ad before you've spent a dollar.
What a real campaign actually looks like
Beyond reach, two structural gaps are worth understanding before allocating meaningful budget.
Targeting is contextual, not firmographic. ChatGPT Ads use "context hints," short topic signals rather than keywords, and the platform matches ads to conversations that align with those signals based on full conversational context. What's missing is everything B2B marketers actually rely on elsewhere:
No job title targeting
No company size or industry filters
No account-based targeting or named account lists
No intent data overlays
No retargeting based on prior ad engagement, though it's reportedly on OpenAI's roadmap
Reporting doesn't yet connect to pipeline. Current reporting shows spend, clicks and basic engagement, but nothing like a search terms report, no visibility into which context hints actually triggered a placement and no native pipeline attribution. CRM integration with Salesforce and HubSpot is on OpenAI's stated roadmap but isn't live yet, which means clean UTM tracking and a manually configured CRM conversion goal are essential before spending anything, since the platform itself won't tell you what worked.
Where things actually stand across AI platforms
| Platform | Ad Status | B2B Relevance Right Now |
|---|---|---|
| ChatGPT | Self-serve since May 2026, no minimum spend | Reaches free/Go tier only, thin volume on niche B2B terms |
| Google AI Mode / Gemini | Ads already running in AI Mode, extending to standalone Gemini app later in 2026 | Strong, connects directly to existing Google Ads accounts and infrastructure |
| Microsoft Copilot | Ads served automatically through Microsoft Advertising | Growing, benefits from existing Bing Ads infrastructure |
| Perplexity | Abandoned advertising entirely in February 2026 over trust concerns | None currently, though its research-heavy audience remains B2B-relevant organically |
| Anthropic Claude | No ads, explicit ad-free positioning | Not applicable, relevant only for organic visibility |
Perplexity tested sponsored follow-up questions and answers through 2024 and 2025 before withdrawing the programme entirely in early 2026, citing user trust concerns, a useful reminder that the current landscape is genuinely unsettled, not a fixed set of options to plan permanently around.
Google's position is the one worth watching most closely for B2B specifically. Because AI Mode ads connect to existing Google Ads accounts, a business already running Search campaigns can extend into AI-generated results without rebuilding campaign infrastructure from scratch, a meaningfully lower barrier than ChatGPT's separate, self-contained ad system.
The market isn't waiting, even with these limitations
Despite the reach and targeting gaps, momentum is real.
ChatGPT Ads expanded beyond the US to Canada, Australia and New Zealand in March 2026, then launched in the UK in June, with major holding companies including Omnicom, WPP and Publicis reportedly already testing early formats and self-serve tools ahead of a wider rollout.
That's nothing; it suggests the industry expects this to matter eventually, even while the current version has real limitations for B2B specifically.
Ad density also varies significantly by industry, so it's worth knowing if you're benchmarking your results against a general figure. Shopping-related queries return a sponsored placement more often than any other query type, and within that, Finance and Insurance see the highest ad density of any sector, while Healthcare sees the lowest.
If your category sits closer to Healthcare's end of that range, expect even thinner inventory than the general figures suggest.
What a sensible first test actually looks like
Rather than mirroring an existing Google Ads structure, organise campaigns around the actual problems your buyers are discussing, each as its own campaign:
Evaluating a category (for example, "software for X")
Comparing alternatives to a specific named competitor
Researching how to measure or improve a particular outcome
Context hints work more like topic signals than keywords; aim for 10 to 15 per ad group, written as short descriptive phrases rather than exact-match terms. Structure this as a separate, incremental budget line from the outset, not a reallocation from Google or LinkedIn, so a thin result on ChatGPT specifically doesn't get read as a failure of paid media generally.
The bigger shift this sits inside
None of this means AI is irrelevant to B2B buyer journeys, quite the opposite.
Professionals are already turning to ChatGPT and similar tools to research vendors, compare solutions and build shortlists well before they visit a website or speak to sales. The practical implication is that showing up in that research moment matters, whether that's through organic AEO work and understanding how brand visibility actually shows up inside LLM answers, or eventually through paid placement once the platform matures.
Our own take on GEO as the next evolution of SEO and how AI search is actually being discussed at industry events like BrightonSEO covers the organic side of this in more depth. Paid and organic aren't competing priorities here either; the same discipline that builds real search visibility- genuine expertise disclosed honestly rather than generated at scale- is what a future, better-targeted ad platform will eventually reward too.
Why the trust model matters for considered B2B categories
OpenAI has committed to keeping ad placement separate from how the model generates its actual response, a principle sometimes called answer independence, meaning which company paid for a placement doesn't influence what the AI tells the user.
For B2B categories where a buyer is doing genuine due diligence, evaluating security vendors, comparing enterprise software, researching compliance requirements, that separation matters more than it would for a casual consumer purchase.
A buyer who suspects the answer itself is shaped by advertiser payment stops trusting the tool entirely, which is exactly the dynamic that pushed Perplexity to withdraw its own ad programme.
Whether OpenAI holds that line as ad revenue pressure grows is worth watching, but for now it's a genuine point of reassurance rather than a marketing claim to take at face value.
What to actually prepare before testing
A test built without the right groundwork wastes limited inventory instead of learning from it. Before spending anything, get these in place:
A structured summary of your product's actual differentiators, use cases and target segments, written plainly enough that a system matching on context can use it accurately
Messaging that varies by buying context, enterprise versus SMB, technical versus outcome-led, since a single generic pitch performs worse in a conversational format than it does in a static ad
Clean UTM parameters on every destination URL, and a CRM conversion goal that isolates ChatGPT-sourced traffic specifically, since the platform's own reporting won't do this for you
Brand voice and compliance guardrails defined upfront, particularly if you operate in a regulated category, since policy exclusions here can be strict and are still evolving
None of this guarantees meaningful volume; the inventory constraint is real regardless of how well-prepared a campaign is. But it's the difference between a test that tells you something useful about your category and one that just confirms the platform is early, which you already know.
What this means practically
Treat ChatGPT Ads as a small, deliberately time-boxed experiment, not a channel to shift meaningful budget toward yet.
A learning budget in the low thousands monthly, funded as an incremental line item rather than pulled from an existing Google or LinkedIn budget, is enough to learn whether your specific category has real volume on the platform.
Evaluate at 90 days, not 30, since limited inventory means early data is genuinely thin. Build your own tracking infrastructure, UTMs and a dedicated CRM conversion goal before you spend anything, since the platform's native reporting won't tell you what worked.
For most B2B marketers, PPC budget is still better spent where audience access and attribution are proven, and our AI Search Ads service is built around that same honesty, we're upfront that this is early-stage, and we size expectations and budgets accordingly rather than overselling a channel that isn't ready for a B2B audience yet.
The same design-time thinking that shapes why we don't sell CRO as a standalone service applies here too; understanding a channel properly before committing real budget beats chasing a headline number.
-
As a small, time-boxed experiment, yes, worth testing. As a meaningful budget shift away from proven channels, not yet. The reach and targeting limitations are real and structural, not something a well-optimised campaign can work around today.
-
Because ads only serve to Free and Go tier accounts, and B2B buyers, marketing leaders and senior operators in particular, are disproportionately likely to already be on a paid plan. One industry poll found 74% of SaaS and tech marketers were on paid ChatGPT plans, nearly inverting the platform's roughly 81% free-tier population overall.
-
For most businesses already running Google Ads, yes, it's a lower-effort extension of existing infrastructure rather than a separate platform and targeting model to learn from scratch.
-
A modest, incremental line item, in the low thousands monthly, funded separately from existing Google or LinkedIn budgets, evaluated over 90 days rather than 30 given how thin inventory currently is on specific B2B terms.
If you want an honest read on whether AI Search Ads make sense for your business right now, not an oversold pitch, get in touch to speak with David and Sander, our AI advertising leads.
