ChatGPT Ads for B2B: Targeting, Tracking and Whether to Test

Many B2B buyers now open ChatGPT before they open a supplier’s website. They ask it to explain a problem and compare approaches, then arrive at a shortlist already formed. ChatGPT Ads places a labelled sponsored card below the answer in those conversations. Priority Pixels runs AI assistant ads for B2B marketing teams. For a marketing director the practical question is not how the platform works but whether it deserves a share of next quarter’s budget.

ChatGPT Ads for B2B is a different proposition from search advertising, with targeting that works without keywords and tracking limits that matter for long sales cycles. If the platform itself is new to you, our ChatGPT Ads explainer covers the ad format and how ads are selected.

Why B2B Buyers Take Their Research to ChatGPT

B2B buyers use ChatGPT for the exploratory part of a purchase, where they are working out what the problem is and which kinds of supplier could solve it. A conversation lets a buyer describe their situation in full, ask follow up questions and get a comparison back in one place, which suits complex purchases with several stakeholders.

That behaviour is why ChatGPT already matters for organic visibility, as covered in our piece on ChatGPT as a search engine. In its August 2026 announcement on expanding ChatGPT Ads, OpenAI said ChatGPT has more than 1 billion weekly active users and that the ad business reached a US$1 billion annualised revenue run rate in under 200 days.

For advertisers the appeal is timing. An ad below an answer about choosing a managed IT provider reaches someone in the middle of building a shortlist. That is a different moment from LinkedIn Ads for B2B lead generation, where campaigns are built around who the person is, so the two channels tend to sit alongside each other.

How ChatGPT Ads Targeting Works Without Keywords

ChatGPT Ads targets conversations rather than search terms. Each ad group carries context hints, which are plain language descriptions of the conversations, topics or keywords an ad belongs alongside. OpenAI’s guidance on writing context hints is explicit that they are not exact match keywords, do not guarantee delivery and cannot enforce geographic limits, schedules or exclusions.

For B2B that changes how campaigns are planned. Instead of bidding on a phrase, an ad group might describe operations directors comparing ways to replace spreadsheets in a manufacturing business. Priority Pixels writes those descriptions from a client’s real search term data and the questions their buyers ask, one per ad group, then changes them one at a time so the effect of each change can be read.

The remaining controls sit at campaign level. The settings below come from OpenAI’s campaign setup documentation.

The customer list threshold is the setting most B2B advertisers run into first. OpenAI’s custom audiences guidance requires at least 25,000 matched users before a list can be used for inclusion or bid multipliers of 0.1x to 10x, with 100,000 or more recommended. Smaller lists can still exclude existing customers from acquisition ads, which for many B2B firms is the realistic use.

OpenAI says on its ads testing page that it keeps guardrails to prevent narrow ads targeting, so this is not an account based marketing tool. For B2B, the quality of the context hints does most of the work.

The Free and Go Audience Question

ChatGPT Ads only reach people on the Free and Go plans. OpenAI’s help article on ads in ChatGPT confirms that Plus, Pro, Business, Enterprise and Edu accounts do not get ads. Free users can switch to an ad free option in exchange for lower message limits and fewer tools.

This is the main reason B2B advertisers should set expectations carefully. Staff working in a company’s ChatGPT Business or Enterprise account are outside the ad audience, so a campaign reaches buyers researching on a personal Free or Go account. We have not seen OpenAI publish a breakdown of ad reach by plan or industry, so the reachable share for your sector is one of the things a test finds out.

Privacy settings shape the audience further. OpenAI’s privacy policy for the EEA, UK and Switzerland relies on consent for personalised ads and on legitimate interests for contextual ads based on the current chat. Digiday reported that this consent model was added as ChatGPT ads went live in the UK in June 2026. Users who turn personalisation off still see contextual ads based on the current conversation.

Tracking Leads From a ChatGPT Ad to Your CRM

ChatGPT Ads conversion tracking runs through the OpenAI Pixel, the Conversions API or a combination of the two, connected to a data source in Ads Manager. OpenAI’s conversion measurement documentation explains that a click reference called oppref is added to landing page URLs and stored by the Pixel in a first party cookie. The same reference can be passed in Conversions API calls. OpenAI tells advertisers to send conversion data only after giving users clear information and obtaining the consents the law requires.

For UK advertisers that cookie brings consent into play. The ICO’s guidance on how the storage and access rules apply to online advertising says consent applies to any storage and access technology used for advertising purposes, including measuring campaign effectiveness. In practice the Pixel should load only once a visitor has accepted advertising cookies.

Attribution is where B2B advertisers need the most care. OpenAI’s guide to measuring results offers click windows of 7, 14 or 30 days and a view window of 0 or 1 day, with last touch attribution rather than multi touch. Conversions take 24 to 48 hours to appear. A deal that closes three months after the first click falls outside every available window, so a test has to be judged on an earlier event such as a qualified enquiry or a booked demo.

OpenAI Pixel

A website script that records conversion events and stores the click reference. It hashes form details in the browser before sending them.

Conversions API

A server connection that sends conversion events from your own systems. It can carry the click reference for matching.

oppref

OpenAI’s click reference, appended to landing page URLs. It links a later conversion back to the ad click that started the visit.

Modelled conversions

Conversions estimated by OpenAI rather than observed directly. Reported totals may include them where available.

The Conversions API is the more useful route for connecting ad clicks to CRM outcomes. A lead captured by a form can be stored with its oppref value, so a later CRM event such as the lead becoming a sales qualified opportunity can be sent back against that reference within the attribution window. OpenAI named HubSpot as its first CRM partner in its September 2026 advertising update, covering campaign creation, lead capture and performance reporting.

Platform figures and CRM figures will rarely match exactly. Priority Pixels reports observed and modelled conversions separately, cross checks every number against the client’s own analytics and CRM and gives daily visibility between monthly reports through the client reporting dashboard. The shift towards server side measurement covered in our piece on PPC tracking after third party cookies applies here too.

Which B2B Businesses Should Test Now and Which Should Wait

The B2B businesses that suit a test are those whose buyers research before they shortlist, whose category is open to UK advertisers and who can track a lead event reliably. SaaS firms, managed service providers, manufacturers and consultancies often fit, though each still needs a category check against OpenAI’s live ad policies before any work begins.

Some regulated sectors need to wait. OpenAI treats health services, legal services and most financial services as US only or restricted. Its onboarding and policy troubleshooting page lists telehealth, online pharmacies and legal services as not allowed outside the US. Deposit accounts, credit cards, insurance, mortgages and investing are limited to select approved advertisers outside the US. Health software and financial software are allowed, so a software vendor selling into healthcare is in a different position from a clinic advertising its own services.

A business on the wait side is not ruled out for good. The ad policies have been revised several times during the beta, so the eligibility check should be repeated rather than treated as a one off. For businesses that are not ready, Google Ads and Microsoft Ads remain the established ways to reach buyers at the point of search.

Structuring a Test That Ends in a Clear Decision

A useful ChatGPT Ads test has a success measure agreed before any money is spent and a fixed read period long enough to judge it. OpenAI’s advertiser FAQ says no cross advertiser performance benchmarks exist yet, so there is no industry figure to compare against. The measure has to come from your own numbers, such as cost per qualified lead on your existing search campaigns.

Budget follows from that measure rather than the other way round. It needs to be large enough to run a proper three month test rather than a token one, with the figure set in pounds and the dollar equivalent recorded. Our piece on what ChatGPT Ads cost covers how bidding and billing work. The stages below show how a test is run so that ChatGPT Ads results can be read with confidence at the end.

The advertiser account should belong to you throughout. Priority Pixels has clients create the account under their own business details and add our team as a member, which keeps ownership clear whatever the test concludes. If you are weighing up who should run a test, our guide to choosing a ChatGPT Ads agency sets out the questions worth asking before you commit.