In May 2026, OpenAI opened ChatGPT Ads to any US marketer with a credit card. The pitch writes itself: hundreds of millions of people reasoning through real problems in real time, and now you can put an offer in front of them mid-thought. Your feed is already full of people telling you to move fast.
Here's what most of them are skipping. ChatGPT hands you something close to the strongest buying signal in advertising - a person actively working through a decision - and almost nothing about who that person is. No job title. No company size. No seniority. You know what someone is thinking about. You have no idea whether they sit on the buying committee you're trying to reach.
For B2B, that one trade-off decides everything. It's the difference between a channel that quietly outperforms LinkedIn for the next twelve months and a line item you'll be explaining away in your next budget review with nothing to show for it.
You can advertise on ChatGPT today. Whether you should comes down to four specific questions about your program — and most teams committing budget right now haven't asked a single one of them.
ChatGPT Ads appear as contextually integrated placements inside or alongside conversational responses. The format sits closer to search than display: less visual friction, but no native lead gen form and no rich media unit. You don't get the creative flexibility of a display unit or the direct response machinery of a LinkedIn lead gen form. You get an offer woven into a moment of reasoning.
That moment is the whole point. People on ChatGPT aren't scrolling. They're working through a problem, asking for structured analysis, looking for a recommendation they can act on. The engagement is self-selected, which raises the stakes on offer design rather than lowering them. A weak offer gets ignored faster here, not slower.
It's also early. Targeting options, ad formats, and measurement tools that don't exist today could ship within the year. The platform will evolve over time.
ChatGPT Ads targets by context and topic. Ads serve when a user's query relates to a relevant subject area. What it doesn't offer is firmographic targeting - no job title filter, no company size parameter, no industry selector, no seniority level.
How much that hurts depends entirely on how your campaigns are built. Make it concrete: a campaign aimed at CFOs at mid-market SaaS companies can't run on ChatGPT Ads today. Neither can VP-level operations leaders at distribution businesses, or IT directors at healthcare orgs over 500 employees. These are standard B2B audiences. LinkedIn's builder approximates all of them with reasonable precision. ChatGPT Ads has no equivalent.
Attribute precision is exactly what makes LinkedIn worth its CPMs for B2B. Google offers keyword intent as a proxy for commercial interest. ChatGPT Ads offers something else: rich query context that tells you what someone is wrestling with, and nothing about who they are professionally. A user asking sharp questions about ERP selection criteria might be a COO, a systems analyst, or a consultant researching for a client. The platform can't tell them apart, and that distinction is load-bearing for most B2B qualification logic.
The strongest fit is broad-ICP awareness. A B2B marketing agency whose target is essentially any company actively investing in digital marketing can credibly reach buyers researching agency selection on ChatGPT. The contextual match is strong without needing firmographic precision. Someone working through how to evaluate an agency is making a relevant buying decision regardless of their exact title.
Thought leadership and decision-support content fit the format unusually well. Frameworks, comparison guides, evaluation tools, educational resources - these match how people actually use ChatGPT. The platform rewards offers that help a user think, not offers that demand an immediate conversion. A guide to evaluating paid media ROI behaves very differently here than a demo request, and that difference should shape what you promote before it shapes whether you promote.
There's a real first-mover case, and it's specific to broad ICPs. Early advertisers on new platforms typically get lower CPMs, more prominent placement, and room to build institutional knowledge before the field fills up. Teams that build playbooks now, while the channel is uncrowded, accumulate learning at a lower cost per insight. This only holds for teams whose ICP is broad enough to benefit from awareness-level exposure. For narrow-ICP programs, early experiments without targeting precision produce noisy data that stays hard to optimize even after the platform adds firmographic controls.
One more factor worth weighing: ChatGPT's user base currently skews toward technically sophisticated, innovation-forward roles. Product, engineering, and marketing leadership adopted AI research tools earlier than procurement, finance, or legal. If your ICP includes those early-adopter roles, the current audience composition is an advantage. If your buyers live in procurement or finance, the skew works against you for now.
Account-based programs are a clear mismatch. ABM depends on account attributes - company size, industry, tech stack, named account lists. Without company-level filters or CRM matching, running ABM on ChatGPT Ads means buying impressions against a broad contextual audience and hoping your target accounts are somewhere in the mix. That's awareness advertising. Calling it ABM doesn't change what the spend is doing.
Narrow-ICP campaigns where title or seniority is load-bearing hit the same wall. Selling CFO advisory means reaching CFOs. Selling procurement software means reaching procurement leaders. On LinkedIn you can build those audiences. On ChatGPT Ads, query context is the only proxy for professional identity, and that proxy carries too much noise for high-precision programs.
Direct response formats face a compounding problem on top of the targeting gap. Demo requests, free trials, and conversion-heavy CTAs require a level of intent and trust that a contextual, search-adjacent placement is unlikely to generate at scale - especially when you can't confirm professional qualification. The format suits research-stage offers. Run direct conversion campaigns here and you'll mostly generate cost data that misleads.
Attribution is the practical constraint, and how much it bites depends on how budget decisions get made at your company. ChatGPT Ads' measurement isn't at parity with Google or LinkedIn, where multi-touch attribution, conversion tracking, and CRM integrations are mature. If you run closed-loop reporting - every dollar tracing to pipeline or revenue to survive internal review - the measurement gap makes the spend genuinely hard to defend after a test. That's a real limit on whether test results can support your next budget decision.
When firmographic targeting arrives, CPMs and competition will rise at the same time. Teams that wait will be building playbooks from scratch, at higher cost, against more competitors. That doesn't mean everyone should test now. It does mean waiting carries a cost that belongs in the decision.
Four questions. Work them in order. Each one narrows the call.
1. How broad is your ICP? If reaching your buyer requires job title, company size, seniority, or industry precision, ChatGPT Ads can't deliver it today. For most programs in this category, the right move is a 12-month roadmap review - revisit when targeting matures or first-party matching is confirmed. If contextual topic-matching can plausibly find your buyers because they're identifiable by what they're researching, move on.
2. What is this spend actually for? Awareness and research-stage content distribution match what the platform does today. Pipeline generation with a measurable cost per qualified opportunity doesn't for two reasons that compound: the targeting precision to reach qualified buyers isn't there, and the attribution to measure conversions isn't mature. If your success metric is cost per qualified opportunity or sourced pipeline, the channel is premature. If it's brand presence and reach among a broad relevant audience, it's a reasonable fit.
3. Can your test budget survive without attributed pipeline? Attribution ambiguity is workable if your org accepts brand signals, qualified site visits, and content engagement as proxies for a test period. It's an operational problem if your budget line requires direct pipeline attribution to survive a review. Define what success looks like before you commit spend, not after a test produces ambiguous results you have to explain in hindsight.
4. Are your current channels showing diminishing returns? The first-mover case is strongest for teams facing rising CPMs or saturation on LinkedIn or Google. If your existing channels are still performing efficiently, the urgency drops and the risk of diffusing budget is real. If costs are climbing and reach is flattening, the case for building presence in a newer, cheaper channel gets stronger.
If your first three answers are yes and your established channels are saturating, a minimal test is a defensible call. If two or more answers point to "not yet," put ChatGPT Ads on the roadmap and revisit when targeting develops.
A few parameters separate a useful first test from an expensive one that produces noise.
Offer. Lead with informational content - a framework, a guide, a comparison resource, a decision tool. These match the research-mode mindset and don't require professional qualification to perform. Skip demo requests and free trials as your primary conversion goal. The context rewards offers that help buyers think, not offers that demand commitment.
Budget. Set a ceiling and treat it as a learning investment, not a performance budget with CAC targets. No reliable CPM or CPC benchmarks exist for ChatGPT Ads yet. Any planning figure is directional. Setting efficiency targets against benchmarks that don't exist produces confident, wrong conclusions.
Measurement. Track qualified site visits, time on page, content engagement, and branded search lift as your primary signals. If your first-party data allows, look at the job titles of post-click visitors to gauge whether contextual targeting is reaching the right people. Accept that direct pipeline attribution will be limited near-term, and design your internal reporting around that before the test starts.
Duration. Plan for at least one to two months of data before drawing conclusions. The feedback loop on a lower-volume, newer channel is slower. Early signals mislead in both directions, and cutting a test short defeats the point of running it.
For broad-ICP teams with discretionary budget, awareness-stage objectives, and tolerance for attribution ambiguity, a minimal test is a reasonable call. The user base is growing, CPMs are still low, and the playbook you build now will be cheaper than the one you build later against more competition.
For teams running account-based programs, narrow-ICP campaigns, or anything that needs clean pipeline attribution to renew budget, the channel isn't ready for your use case yet. It's developing fast - targeting that doesn't exist in mid-2026 could land within 12 to 18 months - so dismissing it entirely is its own risk. Teams that write it off now may end up building from scratch right when targeting matures and the channel gets crowded.
Start with the four questions. Work through them before you commit a dollar, and you'll have a clearer answer than most teams having this exact conversation right now.
If your team is working through this decision, let's talk.
Jake Finkelstein is the Founder and CEO of 10cubed, a Durham, NC-based digital marketing agency helping B2B companies grow through strategy, AI, and automation. A veteran B2B marketer and demand generation specialist, he has spent more than 20 years helping growth-stage and enterprise brands build pipeline, drive revenue, and operationalize modern marketing programs.
