OpenAI turned ads on inside ChatGPT earlier this year. The $200K minimum spend requirement is the first honest signal about who this channel is for.

Not mid-market. Not SMB. Enterprise B2B teams with budget to purchase a learning period while attribution is broken and CTR benchmarks are still forming. That is the actual audience for this channel right now, and being clear about it saves everyone else from making an expensive mistake.

Here is what the Ads Manager actually is, what you are paying for, where the measurement gaps will cost you, and what to do if you qualify.

The Ads Manager Is Real. The Reporting Is Not.

OpenAI confirmed it is testing a self-serve campaign management dashboard with a small group of early partners. The dashboard lets marketers launch, monitor, and optimize campaigns in real time. Targeting is contextual: ads are matched to conversations based on the semantic content of what users are asking, not keywords, not job titles, not company size.

Reporting, as of this writing, is CSV-only. You get impressions and clicks. That is the entire measurement stack. No attribution, no conversion tracking, no pipeline view, no CRM connection. OpenAI has stated that more advanced analytics are coming, but no public timeline exists.

For context on what this means in practice: you will know your ad appeared and whether someone clicked it. You will not know whether that person became a pipeline opportunity, booked a demo, or was even in your ICP. The Ads Manager gives you media delivery data. It does not give you revenue data.

If you are running B2B campaigns today, you already understand the gap between those two things.

The CPM Is Expensive. The Intent Argument Is Real. But So Is the Volume Problem.

Early reports from test partners put CPMs around $60. For comparison: Meta B2B placements typically run $6-23 depending on targeting and placement, LinkedIn B2B targeting runs $30-80 depending on audience specificity, and Google Display lands between $2 and $10. ChatGPT is pricing at the high end of that range based on early data, though benchmarks will shift as the channel scales.

The argument for paying the premium is intent. ChatGPT users running commercially oriented queries are actively problem-solving, often mid-evaluation. A buyer asking "what is the best enterprise data security software for a 500-person SaaS company" is in a buying frame that a LinkedIn scroll does not replicate.

The counterargument is inventory. Early estimates suggest a small fraction of ChatGPT queries are commercially oriented, far below the commercially oriented share of Google Search queries. The universe of high-intent moments is narrow. The CPM premium partly compensates by pricing each impression as a high-value placement. Whether that is justified for your specific ICP is the thing the test has to answer.

One more variable: OpenAI's incentive structure. Per trade press reporting earlier this year, internal OpenAI projections show significant ad revenue targets for 2026 and beyond. The $200K minimum commitment is partly quality control and partly revenue generation from a small cohort of serious advertisers. Treat platform-reported performance data accordingly. OpenAI wants this channel to look successful.

The CTR Gap Is the Real Performance Problem

Early independent research puts ChatGPT's CTR at approximately 1.3%. Non-brand B2B search campaigns typically run 3-5% CTR depending on keyword intent and position.

That gap is real, and it is structural. ChatGPT keeps users in-app. When the model provides a complete answer, clicking an ad link requires the user to leave a conversation they are already in. Google Search is built to send users somewhere. ChatGPT is built to provide the answer itself.

For B2B advertisers, this means impressions do not translate to clicks at the rate you are used to from search. If your campaign math is built on non-brand search CTR norms, ChatGPT numbers will look broken. They are not broken. They are a different channel with different behavior.

The implication: ChatGPT ads are not a search replacement. They are a brand presence play in a high-intent moment. The click may not happen immediately. The brand impression during active vendor evaluation may show up later in a different channel, a later direct visit, or a sales conversation where the buyer mentions they had already heard of you. That makes this a harder attribution problem than search, and the current CSV-only reporting does not help you solve it.

Attribution Is Specifically Broken for B2B

Here is what makes the attribution gap more acute for B2B than the headline numbers suggest.

B2B sales cycles are long. A buyer sees your ChatGPT ad today, does not click, returns to your category comparison via Google three weeks later, attends a webinar in week six, and converts through a sales outreach in week ten. OpenAI's reporting will show an impression from week one. Your CRM will show the sales outreach from week ten. Nothing in between connects.

This is the dark funnel problem that already makes LinkedIn brand advertising hard to attribute. ChatGPT ads land in the same category: you are buying influence in early-stage evaluation moments that do not produce immediate, measurable pipeline events.

That does not make the channel worthless. It means the measurement standard has to change. Self-reported attribution, asking "how did you first hear about us?" on your demo request or contact form, becomes more important. First-touch analysis for your ICP segments starts to matter. Qualitative pipeline interviews with sales capture signals that your attribution tools miss.

If your team has not built this kind of measurement infrastructure already, the ChatGPT attribution gap will make the channel look like it does not work. It may work. With current tooling, you will not be able to prove it either way.

OpenAI has confirmed a Criteo partnership and is in reported talks with The Trade Desk. CRM integration with Salesforce, HubSpot, and Dynamics is on the roadmap. When that ships, the attribution gap shrinks significantly and the cost-to-verify ratio improves. That is the version of this channel that is viable for a broader set of B2B advertisers. It is not this version.

For how the same attribution and signal problem played out when Meta embedded AI in Ads Manager earlier this year, see Manus AI in Meta Ads Manager.

$200K Minimum Tells You Exactly Who Should Be in Market Now

Enterprise B2B teams with three things: budget flexibility that absorbs a $200K test with unclear attribution, a defined ICP that demonstrably uses ChatGPT for vendor research, and organizational tolerance for presenting incomplete pipeline data to stakeholders while the measurement stack is still early.

Mid-market teams spending $10-50K per month on paid media should not be evaluating this channel right now. The minimum commitment alone prices them out, and the opportunity cost of $200K against undefined attribution is too high at that budget scale.

SMB should not be evaluating this at all. The channel is not designed for them, the contextual targeting does not filter by company size or job title, and the return on attention required to learn a primitive self-serve platform is not worth it when LinkedIn and Meta offer more reliable pipeline data at lower minimums.

For enterprise teams that qualify: the market timing argument is real. ChatGPT advertising is in its early adopter phase, which means auction competition is lower than it will be in 12-18 months. The $60 CPM will not stay at $60 when enterprise demand increases. Google Search CPM also started low before auction competition normalized pricing upward. Getting account-level learning and creative testing data now, before the bidding environment prices in, is worth the test cost if the budget exists and the risk tolerance is there.

The Test Worth Running If You Qualify

The test has one job: tell you whether CPL from ChatGPT is close enough to your LinkedIn baseline to justify staying in the channel.

Request early access through OpenAI's partner program. Build the campaign around bottom-funnel, direct conversion: a demo request, a trial signup, or a gated asset tied to a specific capability your ICP would research in ChatGPT. Use contextual targeting focused on problem-specific query categories relevant to your product (vendor evaluation, capability comparison, category research). Set the flight at 30 days.

At day 30, pull CPL from ChatGPT against your current LinkedIn average CPL. The pass metric: ChatGPT CPL within 150% of your LinkedIn CPL. That is the threshold for whether the intent premium justifies the cost premium.

If you pass, the channel earns a sustained budget allocation. If CPL is trending toward 150% but not there yet, extend the test 30 more days and check trajectory. If CPL is landing at 3-4x LinkedIn with no improvement trend, the channel is not ready for your ICP and the right decision is to re-evaluate in Q3 when CRM integration closes the attribution gap.

One parallel action before you launch: add a "how did you first hear about us?" field to your demo request form if it is not already there. Self-reported attribution is going to be the only mechanism for closing the loop on ChatGPT impressions that did not produce a direct click. You want that data before the test runs, not after you are trying to explain results to a skeptical stakeholder.

Not Enterprise Yet? One Variable Changes When This Channel Opens to You.

Watch the CRM integration timeline. When OpenAI ships closed-loop attribution through Salesforce, HubSpot, and Dynamics, the attribution problem shrinks and the channel becomes viable for a wider range of B2B advertisers at lower minimums. That is the version of ChatGPT advertising worth buying.

The other thing worth doing before that happens: document how your ICP uses ChatGPT. What problems do they solve in the tool? What query categories lead to vendor research? What product categories does your ICP compare in conversational AI? This is the targeting intelligence that will determine whether your ICP's ChatGPT behavior overlaps with commercially oriented query contexts. If your buyers do not use ChatGPT for vendor evaluation, the contextual intent argument does not apply to you regardless of how attribution improves.

Build the audience understanding now. The channel will open to more B2B teams in the next 12-18 months. Teams that know their ICP's ChatGPT behavior before it does will be faster to activate and better positioned to optimize when the minimums drop.

ChatGPT ads are live. The CPM is expensive based on early reports, CTR trails non-brand search, and attribution is a CSV file with two columns. That is the state of the channel in March 2026.

For enterprise B2B teams with the budget to test a channel before the market prices it in, that is an early mover position worth taking.

For everyone else: document your ICP's ChatGPT behavior, watch for the CRM integration announcement, and wait until attribution is better before you fully buy in.

New channels always look expensive before they get competitive. ChatGPT will not stay at $200K minimum spend forever.

Paid Signal covers AI's real impact on B2B paid media. Practitioner-first, revenue-focused, no hype.

Subscribe free →