Growth marketer operator · San Jose, CA | ZuAI: 10K → 2M users at $0.02 CAC | $300k/mo ad spend managed
Ar.Bhavesh Pro
sep 1, 2026 chatgpt adspaid ads

ChatGPT Ads for beginners: the three objectives and what each one is for

how ChatGPT Ads actually works, what CPM, CPC and oCPC each do, and who sees your ads. written from three months of running them.

the short version: ChatGPT Ads has three objectives, they are not interchangeable, and picking the wrong one is the most expensive beginner mistake because you cannot convert a campaign from one to another after the fact.

I have been running ChatGPT Ads for a US startup for about three months. India self-serve opens on 4 september, which means a lot of people are about to run their first campaign. this is what I wish someone had told me before mine.

What ChatGPT Ads actually is

OpenAI started showing ads inside ChatGPT on 9 february 2026 in the US, and opened a self-serve Ads Manager on 5 may. it has since expanded to nine countries, then 31 european markets in august, and ads went live in India on 27 august with more than 50 brands through WPP and Omnicom, with self-serve following on 4 september.

two things matter about who sees them. ads appear for logged-in adults on the Free and Go tiers only. Plus, Pro, Business, Enterprise and Education users see nothing. so your audience skews toward people who have not paid for ChatGPT, which is most of them, but it is not everyone.

the placements are sponsored results inside search-style answers, and ads inside the conversation itself when the topic is relevant.

The targeting model is genuinely different

this is the part that trips up people coming from Meta. targeting is driven by the topic of the current conversation, the user’s past chats and memory, and their previous ad interactions. it is not third party cross-web tracking.

practically, that means you are not buying an audience segment, you are buying a moment. someone asking how to fix a specific problem is a different buyer from someone who once visited a competitor’s website, and the intent is usually much higher.

it also means your creative has to make sense next to an answer. an ad that would work in a feed, where the user is bored, often fails here, where the user is mid-task and impatient.

The three objectives

CPM, for reach. you pay per thousand impressions. use it when you are trying to be seen rather than clicked, which for most seed stage companies is almost never. reach is what you buy when you already have demand and want more people to know your name.

CPC, for clicks and traffic. you pay per click and the system optimises for getting them. this is where most people should start, because it gets you real click data quickly without needing tracking to be perfect first.

oCPC, for conversions. you still pay per click, but delivery is optimised toward a conversion event you define. this is what you eventually want, and it is the one with a hard prerequisite: it needs working conversion tracking before it can work at all.

the trap: you cannot convert an existing CPM or CPC campaign into oCPC. you have to create a new campaign with the objective set to conversions from the start. plan for that rather than discovering it after you have built up campaign history you would rather keep.

The order I would run them in

set up conversion tracking first, before you create any campaign. then run CPC to learn what people click. then, once you have enough conversion volume to be worth optimising toward, build a fresh oCPC campaign.

doing it the other way round, launching oCPC on day one with a pixel you installed that morning, is how you get a campaign that spends without learning. under oCPC, broken tracking does not just make your dashboard wrong, it degrades delivery, because the system is optimising against signals that are not arriving.

I have written the tracking setup up separately in ChatGPT Ads conversion tracking, and it is worth doing before you spend anything.

What to budget

India self-serve starts at daily budgets around ₹725, which is low enough that the real constraint is not money, it is patience. any channel needs enough volume before the numbers mean something.

before you launch, work out what a user can actually be worth to you. the break-even CAC calculator gives you the ceiling you cannot bid above, and the ad funnel calculator turns a CPM and a click rate into a real cost per acquisition so you can sanity check the plan before you spend.

Where this fits with everything else

ChatGPT Ads is one channel, and channels decay. it is unusually cheap right now because it is new and most advertisers are not on it, and that will not last. treat it as a window rather than a permanent advantage.

if you want the wider system this sits inside, the paid ads playbook covers the creative and testing side, and the full playbook library has the other fifteen channels.

Sources: launch and objective details from Choice OMG’s field guide and OpenAI’s help centre on conversion-optimized campaigns; India rollout from Storyboard18.

Want this applied to your startup?

the breakdown call is free. bring one real growth problem, leave with the fix.

Let's talk Growth