Do ChatGPT ads work for B2B? The early read

The short answer

For most B2B advertisers in 2026, ChatGPT ads are a test line, not a channel. The mechanics are genuinely interesting — placement matched to the meaning of a live research conversation is exactly where considered purchases happen — but three facts keep the honest recommendation small: the buyers you most want may not be in the audience at all, the early costs on competitive terms are high, and the platform cannot yet prove what the spend did.

None of that says never. It says: capped budget, your own measurement, a written hypothesis, and no core demand-generation money until the surface can carry it. Here is the reasoning, so the decision is yours rather than a vendor’s.

The audience problem comes first

Every B2B media decision starts with whether the buyer is in the room, and this is where ChatGPT ads carry their open question: ads serve only to free-tier and Go-tier users. Paying subscribers see none.

For consumer brands that trade-off is manageable — the free tier is vast. For B2B it cuts against the campaign’s whole logic, because seniority correlates with paying for tools. The founders, executives and budget holders a pipeline needs are disproportionately likely to be subscribers sitting outside the inventory, while the reachable audience skews toward whoever has not paid. There is no published data yet on how much genuine B2B buying intent lives on the free tier. Until there is, “the right person at the right moment” is a promise the inventory cannot verify, and audience fit has to be judged brief by brief. The full eligibility picture — tiers, countries and categories — is in our eligibility guide.

What the early numbers actually say for B2B

The one widely shared public B2B test is worth reading precisely because it is so inconclusive: around $1,000 on competitive B2B terms returned roughly $13 average cost per click, a click-through rate near 1%, a CPM around $120, and no recorded conversions across about 90 clicks — with the advertiser’s own tracking partly misconfigured, so even the clicks were muddy. A single small account on expensive terms proves nothing in either direction, and we read it as a caution, not a verdict; the fuller analysis is in what the early ChatGPT Ads data shows.

The structural read matters more than the numbers. The platform bills on clicks, conversion-based bidding has been slow to arrive, and reporting stops at impressions, clicks, CPC and CPM. For a discipline built on cost-per-opportunity and pipeline attribution, that means the burden of proof sits entirely on your own instrumentation. A B2B team that cannot measure the channel itself should not be in it yet.

When a B2B test is worth running

A capped test earns its place when a few things line up. Your buyers plausibly research in assistants and at least part of the committee sits on the free tier — true more often for practitioner-led purchases than for board-level ones. Your category is eligible and your terms are not so contested that the auction starts at painful prices. And you can afford the tuition: the honest frame for 2026 spend is R&D, the same way early paid-search and paid-social advertisers bought years of familiarity before the proof arrived.

Structure it like the experiment it is. A written hypothesis and a defined read-out date. A budget you are prepared to lose. Benefit-led creative in several variations, each mapped to a real question in the buying journey — the mechanics of why breadth wins are in how advertising inside ChatGPT works. And your own measurement layer, UTM-disciplined and conversion-tracked on the pages that matter, so the decision at the read-out rests on evidence.

The stronger B2B play on the same surface

Here is the strategic point the ad platform’s novelty obscures: the same B2B buyers you would pay to reach beside the answer are already reading the answer itself — and the sources it cites. Being one of those cited sources is not gated by ad policy, is not confined to the free tier, and compounds instead of ending when a flight does. Paid subscribers who never see an ad still see the citations.

That is the organic discipline of AI search visibility, and for most B2B brands it is the higher-conviction investment on this surface today: measurable at predictable cost, durable across model updates, and reaching the exact audience the ad inventory cannot. The mature posture is not either/or. It is organic first as the foundation, a capped paid test as the complement, and both measured in one frame — which is how we run ChatGPT advertising as a managed channel inside a connected performance programme.

The short version

B2B buyers are researching inside ChatGPT, but the ad inventory cannot yet prove it reaches the ones who sign. Free-tier-only serving skews the audience away from senior buyers, early clicks on competitive terms are expensive, and the platform’s reporting leaves proof of return to your own analytics. Run a capped, instrumented test if the audience logic holds for your category — and put the conviction budget into being the source the assistant cites, which reaches every tier and keeps working after the spend stops. If you want the test designed and the organic layer built as one programme, start here.