• Merito
  • Posts
  • ChatGPT Became a Salesperson

ChatGPT Became a Salesperson

It's Taking Bookings for ₹725 a Day

D2C x India

Hey readers,

Welcome to the thirtieth edition of D2C Cents!

TLDR - We are your scroll-friendly, no-fluff download of what's shaping India's D2C brands.

For three years, ChatGPT was one corner of the internet that didn't feel like it was trying to sell you something. You asked a question, and it answered with no sponsored clutter. That was a big reason people trusted it so quickly. 

As one media veteran put it, "the reason quick adoption happened with LLMs was also that they were coming without brand endorsements."

More than 50 brands are going live in India this launch week through WPP and Omnicom, OpenAI's first agency partners here, though the roster hasn't been made public. 

India is OpenAI's second-largest market, with 100M+ weekly users. OpenAI began testing ads in the US back in February 2026, and the early signals there show how fast this is scaling. In the US, just over half of all free-tier replies already carry a sponsored block. The pilot reportedly reached ~$100M in annualised revenue within its first six weeks.

The ads are already a serious business.

ChatGPT's ads work differently from Google and Meta, especially in how users are matched to them.

OpenAI's own example ad is a "Heirloom Groceries" sponsored product card that appears below a ChatGPT answer, labelled and visually separated

  • Where they appear: a single block below ChatGPT's answer, with the advertiser's name, logo, headline, description and image, clearly labelled "Sponsored" and visually separated, plus a "why am I seeing this" disclosure. During the test, they don't appear in the ChatGPT Atlas browser or in Temporary Chats.

  • Who sees them: only logged-in Free and Go users aged 18+. Plus, Pro, Business, Enterprise, and Edu users remain ad-free. That means advertisers are reaching the more price-sensitive, discovery-stage majority, and a young one: nearly half of ChatGPT messages come from people under 26.

  • How many you'll see: a unit can show one or more items, and crucially, it "may feature one or more items from an advertiser, or multiple advertisers." So you may not just be sitting below a competitor; you may be sitting inside the same ad unit as one. The shortlist logic applies to the ads too. 

  • How you're matched: there are no keywords. OpenAI looks at the context and intent of the live conversation, along with your landing page, ad copy, and the "context hints" you provide about the topics where your business is relevant. If a user has ads personalisation turned on, OpenAI can also use their broader ChatGPT history and memory.

  • How you pay: you're charged per click or per impression, through a relevance-weighted auction, and you set a daily budget. India self-serve opened September 4, with daily budgets starting at ₹725.

You can ask ChatGPT about an ad. Tap the ad's menu, hit "Ask ChatGPT," and it'll answer questions about the product, a behaviour no other ad platform has, and one that rewards brands whose data actually holds up to a follow-up question.

Ads are blocked near sensitive topics (personal health, mental health, politics) and in Temporary Chats. If your category touches those, the surface simply won't serve you.

That's the whole game in three sentences. You can pay for the sponsored card, but you cannot buy your way into what ChatGPT actually says. Which is exactly why the organic answer, not the ad, is the real prize.

The most important part is that the ad is separate from the answer. OpenAI calls this "answer independence." Paying for an ad cannot change, rank, or influence what ChatGPT says about your category.

That distinction matters because it means buying visibility and earning a recommendation are two different things.

Most advertising interrupts. You're scrolling Instagram, an ad appears, and you resent it. ChatGPT ads appear at a different moment, when someone is already trying to decide what to buy. 

"Best stroller under ₹15,000 for a small car.
"Which protein powder for a beginner?

The person has already described exactly what they need. The ad is showing up in the middle of that decision, rather than interrupting something else.

The shelf is also much smaller. Google gives you ten links and a wall of options; ChatGPT gives you three to eight products, with a reason for each.

The early numbers reflect that.

Traffic is converting 1.5x–3x higher than search. 

In the US, CPCs aren't cheap at $3–18, but the intent behind each click is much higher. 

Retail, grocery, travel, and B2B software moved first, with Best Buy and Lowe's among the named early advertisers.

There's also a creative lesson emerging that feature lists underperform. Because the user is mid-decision and can just ask ChatGPT a follow-up, advertisers are finding that leading with the problem you solve, and with proof (reviews, guarantees), works better than listing specs.

Because the ad and the answer are separate, "just run ads" isn't enough. You could pay to appear directly below a response that recommends your competitor by name. The ad can build on relevance you've already earned; it can't create that relevance. So the priorities are:

  1. First, check what ChatGPT already says about you. Run your category prompts and searches like "best [your product] in India" ten times. If it keeps naming your competitors and not you, fix that first. Paying for an ad above a bad organic answer is money wasted.

  2. Fix your organic AI presence (AEO) before you spend. The same signals that help AI recommend your brand organically can also make your ads more relevant. So build the foundation first.

  1. Clean and structure your product data. OpenAI matches ads using your landing page, copy, and context hints. Brands already running Google Shopping or Meta Commerce feeds have a head start. Everyone else should get their product data export-ready and write it in conversational language rather than keyword strings.

  2. Write context hints, not keyword lists. The buying model works around intent and topics. Describe the conversations where your brand belongs instead of listing every possible match term. Keyword targeting doesn't translate directly here.

  3. Start small, on CPC, as a learning budget. ₹725/day and $3–5 test bids are enough to learn how a conversational, decision-stage surface performs for your category. Don't treat that as a scaling budget. Treat it as a Beta experiment, because that's what it is.

  4. Change the metric you judge it by. Reporting gives you impressions, clicks, and conversions, but click volume can be misleading on a bottom-of-funnel surface. Watch downstream conversion and intent quality, and tag your URLs so you can actually attribute the traffic.

The bigger opportunity here is timing.

Right now the channel is new, most of your category hasn't figured it out, and you can still learn cheaply before CPCs climb. Brands that got onto Meta and quick commerce early rode exactly this kind of wave; the cost of learning was low because nobody else had crowded in yet.

This matters because ChatGPT doesn't feel like a feed. Instagram users see "Sponsored" and know how to treat it. ChatGPT feels more like advice from a knowledgeable friend. So when a paid block appears next to that advice, will users treat it like any other ad, or start wondering whether the answer itself was influenced?

The risk is bigger than search. Google has spent decades separating ads from organic results. ChatGPT's appeal is one direct, authoritative answer. If users start doubting whether sponsorship affects those answers, trust in the whole product can suffer.

OpenAI clearly anticipated this concern and addressed it directly.

So far, consumers seem fairly comfortable with it. A May 2026 survey of 1,000 US AI users found 82% considered ads inside AI assistants at least as trustworthy as Google Search ads, and half said they'd research a product further after seeing a relevant one. 

But there's a catch. Gen Z skepticism toward AI ads is rising, according to the IAB, and Gen Z is also one of ChatGPT's biggest user groups.

Nobody knows where this lands yet. For brands, the lesson is simpler: on a platform people trust for advice, a bad ad can cost more than a bad click.

The bottom line

Before you run a single ChatGPT ad, do the harder thing. Become the brand ChatGPT already names when someone asks. 

Fix the product feed, build your AEO presence, and make sure ChatGPT recognises your brand as a relevant option. Then test the ₹725/day ad budget.

Get the order wrong, and you're paying to appear next to a recommendation for your competitor.

See you next edition - same time, deeper insights.

Until then, keep building strategically.
Abhishek