ChatGPT Ads Are Coming: Should Your Shop Test Them?

A coaching-centre owner in Nagpur got a call last week that should sound familiar. An agency told him ChatGPT ads were "live in India now", that his competitors were already buying them, and that he needed to book before Diwali ended. He nearly said yes. Then he asked one question the caller couldn't answer cleanly: show me where India is on the list.

That question saved him money. On Oct 21, 2026, OpenAI expanded ChatGPT ads by 26 countries, to 89 in total — but India is not among the named new markets. So this is not a "run ads today" guide. It's the test plan to build now, so you spend on purpose, not panic, when ads open here.

In this post:

• What ChatGPT ads actually are
• Who sees them, and who never will
• Are ChatGPT ads available in India yet
• ChatGPT ads vs Google Ads
• The one-offer test
• The 3 numbers that decide keep or kill
• A worked example with a kill budget
• Traps in a first campaign
• Questions to ask before you sign anything
• FAQ

What ChatGPT ads actually are

ChatGPT ads are sponsored placements inside ChatGPT answers. A person asks a question, gets an answer, and sees a marked ad alongside it — a product, a service, a link to learn more or buy.

The ad sits in the flow of a conversation, next to the help the person just asked for. That placement is the whole pitch, and the whole risk. Done well, it feels like a timely suggestion. Done badly, it feels like someone barging into a private chat to sell something.

For a chatgpt ads small business test, keep that picture in mind. You're buying a moment when someone is already trying to solve a problem, not eyeballs on a feed. Your ad has to be the next useful step, not a louder shout. OpenAI has built this since a US pilot in February 2026, with cumulative advertisers past 100,000 since then — a sign of curiosity, not proof small shops profit from it yet.

Who sees them, and who never will

Ads show only on the Free and Go plans. Paid tiers — Plus, Pro, Business and Enterprise — stay ad-free.

So the person who sees your ad uses ChatGPT free, or the cheaper Go plan. They may be price-sensitive and early in their search, not yet ready to buy. That's not a bad audience — just a different one from a paying-plan user who expects a clean workspace.

Are ChatGPT ads available in India yet

Short answer: not confirmed for Indian advertisers from the October announcement.

The Oct 21 expansion added 26 countries, taking the total to 89. Named new markets include Argentina, Chile, Colombia, Peru, Nigeria, South Africa, Serbia, Ukraine, Cambodia, Bangladesh and Pakistan. India is not on that named list.

If someone tells you Indian advertisers can buy ChatGPT ads right now, ask for proof: a screenshot from an ads account set to India, a help page naming India, or a dated announcement. A sales call is not proof.

My advice is blunt: test, don't chase. Keep your budget where it already works, and check the official announcement again in a few weeks.

ChatGPT ads vs Google Ads

On Google, a person types "plumber near me". They know what they want and are choosing between options, so your ad competes on price, rating and speed. Inside ChatGPT, a person asking "how do I plan a small shop's Diwali offers?" is still figuring out the problem, so your ad competes on relevance — and on whether it helps before it sells.

	Google Ads	ChatGPT ads
What the user is doing	Searching for a known thing	Talking through a problem
Best moment to appear	When they compare options	When they ask for a next step
What wins	Clear price, proof, speed	A useful answer, then a clear offer
What fails	Vague claims	A hard sell in the middle of advice
Who sees ads	Searchers, broadly	Free and Go plan users only

So don't copy-paste your Google Ads here. "Best coaching in Nagpur, call now" works on a search page but interrupts inside a chat. "Free 20-minute study-plan call" fits, because it offers the next step the person wanted.

The one-offer test

Most small-business ad tests fail for a boring reason: too many offers at once. Three services, two landing pages, one vague "learn more" button — and nobody knows what worked. So your first test is one offer, one landing page, one action.

One offer. Pick the single thing a stranger can understand in ten seconds. "Free trial class this Saturday" beats "quality education for all classes" every time.

One landing page. Not your homepage, with its menu and six buttons. One page, one headline matching the ad, one form or one WhatsApp button. If the visitor has to hunt, you've lost the click you paid for.

One action. Book, call, or message — pick one. A page that asks for all three asks for nothing clearly, and gets nothing.

Then set the kill budget before you spend a rupee: the amount you agree, in advance, to lose while learning. When it runs out, stop and judge — don't top it up because "it might turn around tomorrow". Write it down, so a test can't quietly become a habit.

Don't have a landing page yet? Fix that first — see our guide to finding your first freelance clients without paid ads.

The 3 numbers that decide keep or kill

After the test, you'll be tempted to judge it by clicks. Don't. Clicks are cheap to count and easy to flatter. Judge it by three numbers, in order:

1. Cost per lead. Spend divided by real enquiries — a form filled, a call made, a WhatsApp message sent. Not a click, not a page view.

2. Lead-to-customer rate. Of those enquiries, how many became paying customers? Ten cheap leads that buy nothing lose to two dearer leads that both buy.

3. Cost per paying customer. Spend divided by paying customers, compared with what a customer is worth to you. If a new student pays ₹3,000, and each paying customer cost ₹2,500 in ads, the test is not a win yet — it's a warning.

Run those three in order, every time. If the first looks great but the third looks awful, the ad is attracting the wrong people.

A worked example with a kill budget

The numbers below are an illustrative example, not ChatGPT ad rates. OpenAI has not published a rate card for Indian advertisers, and anyone quoting you a fixed cost per click today is guessing.

Say you set a kill budget of ₹5,000. That's your choice, your cap, written down before you start.

Step	Illustrative example
Kill budget	₹5,000, fixed in advance
Enquiries from the test	25 real messages or forms
Cost per lead	₹5,000 ÷ 25 = ₹200 per lead
Enquiries that buy	5 out of 25
Lead-to-customer rate	5 ÷ 25 = 20%
Cost per paying customer	₹5,000 ÷ 5 = ₹1,000

Judge it against your own business. If each new customer is worth ₹4,000, ₹1,000 to win one leaves room for profit — keep testing, slowly. If each customer is worth ₹800, the same test loses money on every sale — kill it, however exciting 25 enquiries felt.

Traps in a first campaign

Start and end controls may lag. Coverage warns these controls may not appear until after publishing. On day one, check you can pause the campaign and that spending stops when you say stop. If you can't find the pause button, don't fund the account until you can.

Run-rate claims are not revenue. OpenAI's post claims a $1B annualised run rate in under 200 days. A run rate is a pace, not money in the bank. It's a sign of growth, not proof your shop will profit.

"India is live" without proof. Demand a screenshot or an official page. An agency that can't show one is selling urgency, not access.

Questions to ask before you sign anything

Copy this list into your notes app and use it on any agency call:

Before you sign a ChatGPT ads pitch, ask: 1. Show me proof Indian advertisers can buy this today — a screenshot, not a claim. 2. What is the one offer, and what is the one landing page? 3. What is my kill budget, in writing, and who can pause the campaign? 4. How will you report cost per lead, lead-to-customer rate, and cost per paying customer? 5. If the test fails, what do I learn — and what do I get to keep?

The bottom line: ChatGPT ads are real, growing fast, and not yet confirmed for Indian advertisers. That's not a reason to ignore them, and not a reason to pay for them today. Build the one-offer test, set the kill budget, and learn the three numbers. When the channel opens here, you'll be ready, with a plan that protects your money.

Which offer would you test first? Pick the one a stranger understands in ten seconds. That's your test. Everything else can wait.

FAQ

What are ChatGPT ads, and who actually sees them

Sponsored placements inside ChatGPT answers, shown only on the Free and Go plans. Paid-plan users do not see ads.

Are ChatGPT ads available to advertisers in India yet

Not confirmed from the Oct 21, 2026 announcement, which added 26 countries to reach 89 total. India is not among the named new markets. Ask anyone claiming Indian access for proof, such as a screenshot from an ads account set to India.

How is advertising inside ChatGPT different from Google Ads

Google Ads catch people searching for a known thing. ChatGPT ads appear while someone is still talking through a problem, so the ad must fit the conversation and offer a useful next step.

What is a sensible first test budget for a small business

No official rate card exists for Indian advertisers, so no fixed price can be honestly quoted today. Set your own kill budget — an amount agreed in advance to spend while learning, such as ₹5,000 above — and stop when it runs out, then judge by the three numbers.

What 3 numbers tell me whether to keep spending or kill the test

Cost per lead, lead-to-customer rate, and cost per paying customer. Clicks alone don't decide it. If the cost per paying customer is higher than what a customer is worth to you, kill the test, however good the clicks looked.

What traps should I watch for in a first campaign

Check you can pause spending from day one, since start and end controls may lag. Treat the $1B run-rate claim as a pace, not booked revenue. Demand proof of Indian access.