B2B buyers now ask ChatGPT the questions they used to type into Google. How do I automate vendor verification? Which API handles GST checks at scale? When ChatGPT Ads opened to self-serve advertisers in India at the end of August 2026, those moments became buyable for the first time.

Our client, a B2B SaaS company selling GST verification APIs to large enterprises, already ran on proven search and social channels. We proposed a structured test: a real budget, every available objective and clear success criteria. The goal was not to chase a shiny new platform. It was to learn early, with data, whether this channel deserved a place in the media mix.

On Google or Meta, years of benchmarks tell you roughly what good looks like. On ChatGPT Ads there was nothing to compare against: no industry CPCs, no proven campaign structures and a measurement setup still in beta.

The user’s mindset is different too. Someone in the middle of a ChatGPT conversation is solving a problem, not scrolling a feed. So the real question was not whether ChatGPT Ads works. It was which objective, which audience, which message and which bidding approach turn that problem-solving moment into a B2B conversion.

Instead of betting on one campaign type, we ran all three objectives the platform offered, each with a defined job. Reach would test whether visibility alone could find the right audience. Clicks, pointed at a custom audience list, would test whether a known B2B audience responds inside ChatGPT. Conversions, with open targeting, would test how well the algorithm could find buyers on its own.

Inside each campaign we rotated four message angles, from automation to risk to real-time and bulk verification. We also ran a bidding test: manual bids against automated bidding, each with its own creative set.

Reach delivered 14,881 impressions and 358 clicks but no conversions. It wasn’t reaching the decision-makers we needed, so we paused it and moved the budget to the other two campaigns.

The surprise was the Clicks campaign. Aimed at a custom audience, it produced 33 conversions at ₹436 each with a 5.3% CTR. The Conversions campaign, given full freedom to explore, produced 17 conversions at ₹617 each. On a new platform, telling the algorithm who to talk to beat letting it guess.

Four ads ran inside the Clicks campaign. “Automate GSTIN verification,” positioned for large enterprises, earned a 9.76% CTR and 21 conversions at roughly ₹191 each. The risk-led “Identify GSTIN risks” ad took 63% of the spend but converted at around ₹756, nearly four times the cost.

The real-time and bulk-verification angles drew clicks but no conversions. People asking ChatGPT for help respond to the result they want, which is less manual work, not to the problem they’re trying to avoid.

• 01
  The objective is a strategy decision.Reach, Clicks and Conversions behaved like three different channels. Choosing the objective came before choosing the budget.
• 02
  Audience signals matter more on a young platform.A custom audience on the Clicks objective beat open targeting on the Conversions objective by about 30% on cost per conversion.
• 03
  Say what the buyer gets.Outcome-led messaging (“automate”) outperformed risk-led messaging by nearly four times on cost per conversion.
• 04
  Conversions need context.A platform conversion is not yet a qualified lead. Volume only matters once it’s matched against pipeline and revenue.

ChatGPT Ads is still in beta and will keep changing. That is exactly why being early matters. The advertisers who build structured learning now will know what works when the competition arrives.

The next phase moves budget behind the outcome-led message and the custom-audience structure. It brings audience signals into the Conversions objective, separates bidding from creative in a clean test, and connects every conversion to CRM data. Then the channel can be judged on qualified pipeline and revenue next to Google and LinkedIn, not on platform conversions alone.