ChatGPT Ads in India: What OpenAI's Rollout and Razorpay's New Partnership Mean for Shopify D2C Brands

ChatGPT ads in India have been live since late August, and Razorpay began offering to run them for brands on 5 October. What is confirmed, what a small test costs, and how a Shopify D2C brand should prepare.

ChatGPT ads in India went from announcement to agency product in about six weeks. OpenAI began showing ads to logged-in adults on ChatGPT's Free and Go plans in India at the end of August 2026, with a self-serve Ads Manager due to open in September and daily budgets reported to start at ₹725. On 5 October, Razorpay announced that its growth marketing platform, Razorpay Engage, would help Indian brands prepare catalogues for ChatGPT ads, run campaigns and measure them, naming brands including Tanishq, Traya and Palmonas as early participants. For a D2C brand on Shopify, the practical question is not whether AI chat will become an advertising channel. It already is. The question is whether it is worth your next ₹20,000 of test budget, and what has to be true about your store before it is.

We build software for Shopify stores, most of them Indian D2C brands, and we have written before about how to get cited in AI answers without paying for it. This post is about the paid side. It sets out what OpenAI, Razorpay and the press have actually said, flags what we could not confirm, and gives the preparation steps we would follow before spending anything. We have no commercial relationship with OpenAI or Razorpay.

What is confirmed about ChatGPT ads in India

OpenAI's own India announcement page was not accessible to us, so the details below come from TechCrunch's and Inc42's reports of 27 August 2026 and from OpenAI's help article on ads in ChatGPT, as surfaced in search. Where they differ, we say so.

• Who sees ads: logged-in adults on the Free and Go plans in India. Plus, Pro, Business, Enterprise and Education plans stay ad-free, according to Inc42.
• Where they appear: below ChatGPT's answer, when a relevant sponsored product or service is available, clearly labelled and visually separated from the answer, per Inc42.
• Who is excluded: users under 18, and conversations on sensitive topics such as mental health and politics, per Inc42.
• What advertisers get: aggregate campaign data such as impressions and clicks. Advertisers do not get access to conversations, chat history, memories or personal details, and OpenAI says ads will not influence ChatGPT's responses, as reported by Inc42.
• Launch partners: WPP and Omnicom as the first agency partners in India, with more than 50 brands expected to go live within about a week of launch, per TechCrunch and Inc42.
• Self-serve: Inc42 reported that businesses could use the self-serve Ads Manager from 4 September; TechCrunch said it would launch the following month. Both reported a minimum daily budget of ₹725, about $7.60.
• Scale: TechCrunch notes OpenAI said in February 2026 that ChatGPT had more than 100 million weekly active users in India.

OpenAI's help article on ads describes how an ad is chosen: the context and intent of the current conversation, the ad's landing page, title and copy, the advertiser's context hints and targeting selections, and, if the user has ads personalisation switched on, select signals from their wider ChatGPT use. Ranking weighs relevance and advertiser bids. Users can control ads personalisation and clear their ads data. OpenAI notes availability may change as testing expands.

We could not independently confirm that self-serve access opened on 4 September for every Indian advertiser, or what ad formats are available today. Check ads.openai.com and OpenAI's ads documentation for the current position before planning around a specific feature.

What Razorpay announced on 5 October

Razorpay's newsroom release, dated 5 October 2026, says the company is working with OpenAI to help Indian brands advertise on ChatGPT through Razorpay Engage. According to the release, Engage will assess and improve catalogue readiness, structure product feeds, create and manage campaigns, and measure and optimise performance. It says Engage uses an AI agent built around ChatGPT ads best practices to manage campaigns across consumer contexts, with Razorpay's team providing human oversight where judgement is needed. The release names Fastrack, Traya, Tata Neu, Rivaah by Tanishq, Shaadi.com, Axis Max Life, Tanishq, Palmonas and NMIMS as early participants, and includes comments from Razorpay's COO, OpenAI's head of enterprise sales for India and Tanishq's CMO.

Razorpay's ChatGPT ads page describes the offer as early access, with a free website audit for AI readiness and a five-stage process: audit, prepare (structure your catalogue), launch, optimise and scale. It does not publish pricing, minimums, eligibility or specific ad formats.

MediaNama's report of the same week adds context that is not in the release: Razorpay is also working with OpenAI and NPCI on ChatGPT-native storefronts, catalogue syncing and, eventually, in-chat purchases, and about 50 brands and several marketplaces were in agentic-commerce pilots as of August. Neither the release nor that report says whether each named brand's campaign is live, and being listed as a participant is not the same as a published result.

Ads versus citations: two different things

It is worth separating two ways a brand can appear in ChatGPT, because they are easy to confuse and they need different work.

• Organic mentions and citations happen when ChatGPT's answer names or links to your brand because its sources suggest you are relevant. You cannot buy these. OpenAI says ads do not influence answers.
• Ads appear in a separate, labelled unit below the answer, chosen by context, your landing page and copy, targeting and bid. You pay for these, and you can switch them off.

The work overlaps more than it seems. Both reward clear, structured, accurate product information on pages a machine can read. Both punish thin pages, vague titles and missing prices. A brand that has done the work for organic AI visibility, which we covered in our posts on AEO and GEO and on product feed hygiene, starts any ad test with a large advantage.

Who should test ChatGPT ads now, and who should wait

Our view, based on what is confirmed and what is not.

• Good candidates: brands whose products are researched before purchase, such as jewellery, skincare routines, furniture, electronics accessories and gifting, where shoppers describe a need ('a gift for my sister under ₹3,000 who likes minimal jewellery') before naming a product. These are the moments OpenAI's own description of ad selection is built around.
• Good candidates: brands with clean catalogues, honest pricing, fast product pages and a clear return policy. An ad sends a shopper to a landing page; if that page disappoints, the ad budget is wasted.
• Wait: brands in health-adjacent or regulated categories, until you have read OpenAI's ad policies for your category. Sensitive topics are excluded, and you do not want to discover that through rejected ads.
• Wait: brands whose own measurement is not ready. If you cannot tell which orders came from which channel today, a new channel will only add confusion.
• Wait: brands with no spare budget. A test that cannot run for long enough to learn anything is worse than no test.

What a small test costs, as an illustration

Using the reported minimum daily budget of ₹725, a 30-day test at the minimum comes to ₹21,750 before any agency or management fee. That is the arithmetic, not a recommendation; the right budget depends on your order value and what you need to learn. If your average order is ₹1,500, that budget would need about 15 orders just to match its spend in revenue, before product cost, shipping, payment fees and returns. Work out your own break-even before you start, using your contribution margin rather than your price.

If you use a managed service such as Razorpay Engage, ask for the fee structure in writing, what you own at the end (the campaigns, the data, the structured feed), and how results will be reported. Razorpay has not published pricing.

How to prepare a Shopify store for ChatGPT ads

Whatever you decide about spending, these steps are worth doing because they help with organic AI visibility and with Google too. Razorpay's own process starts with the same thing: catalogue readiness.

1. 1Fix product titles. A title should say what the product is, in the words a shopper would use, with the key attribute: '18K gold diamond stud earrings, 0.25 ct' rather than 'Aurora'. Keep brand names and collection names, but not instead of the description.
2. 2Fill in the attributes that answer questions: material, size, colour, weight, ingredients, compatibility. Use Shopify's category metafields where they fit, so the same data feeds your filters, your structured data and any feed.
3. 3Make prices and availability accurate everywhere. Clear stale compare-at prices, and make sure out-of-stock variants are marked correctly. An ad that lands on an unavailable product is money thrown away.
4. 4Check structured data on product pages. Product, Offer and, where honest, review markup help machines read your pages correctly.
5. 5Write collection pages as landing pages. If an ad lands a shopper on 'gold earrings under ₹10,000', that collection should open with in-stock, relevant, well-photographed products and a short description that answers the obvious questions.
6. 6Speed up the page. Shoppers coming from a chat answer have just read a confident recommendation; a slow page is a jarring end to that.

How to measure a ChatGPT ads test properly

Advertisers get aggregate data such as impressions and clicks, according to the reports of OpenAI's launch. Everything after the click is your job, and your Shopify store is where that measurement lives.

1. 1Tag every ad link with UTM parameters, using a consistent source and medium such as source=chatgpt and medium=paid, and a campaign name per test.
2. 2Separate paid from organic. ChatGPT also sends unpaid referral traffic when it links to sites in answers. Without tags, you will credit organic visits to the ads, or the reverse.
3. 3In Shopify Analytics, look at sessions, conversion rate and orders by UTM source and campaign, and compare against the same landing pages' performance from other channels.
4. 4Annotate the start and end of the test in Shopify Analytics, so a later reader knows why traffic changed.
5. 5Judge on contribution margin, not clicks. A channel with expensive clicks can still win if those visitors buy more and return less.
6. 6Run it long enough. One week is rarely enough to see past day-of-week effects, and festive weeks are noisy. Two to four weeks is a more realistic minimum for most stores.

A first ChatGPT ads test, step by step

If you decide to test, keep the first one small and narrow, so that whatever happens, you learn something specific. This is the plan we would use for a typical Shopify D2C brand. It assumes the self-serve Ads Manager is available to you; if you go through a partner, the same structure applies, and you should ask them to follow it.

1. 1Pick one category where shoppers describe a need before they name a product, and where you have at least a few dozen in-stock products. Gifting, a skincare routine or a jewellery category usually fits better than a single hero product.
2. 2Write down the questions a shopper would plausibly ask ChatGPT for that category, in their words, including price ranges and occasions. These shape your ad copy and any context hints the Ads Manager accepts.
3. 3Choose one landing page per question cluster, usually a collection. Make sure each opens with in-stock, relevant products and a short description that answers the question the shopper asked.
4. 4Write plain ad copy that matches the question: what the product is, who it is for, the price range and one honest reason to click. Avoid claims you cannot support; the shopper has just read a careful answer and will notice hype.
5. 5Set a fixed budget and a fixed end date before you start, and decide in advance what result would make you continue: for example, a cost per order below your break-even, measured on contribution margin.
6. 6Change nothing else on those landing pages during the test unless something is broken. If you re-sort or redesign mid-test, note the date with an annotation so you can separate the effects.
7. 7At the end, compare paid ChatGPT traffic with your other paid channels on the same landing pages: conversion rate, order value, returns and cost per order. Then decide to stop, continue at the same level, or widen to a second category.

Two weeks into a test, resist the urge to judge on clicks alone. A conversational shopper may click less often and buy more often than a search shopper, or the reverse. Only your order data will tell you which.

Questions to ask before you hand this to an agency

• Which ad formats are available to us in India today, and where is that documented?
• What will you change on our store or product data, and who owns those changes afterwards?
• How do you separate paid ChatGPT traffic from organic ChatGPT referrals in reporting?
• What is the total cost, including management fees, and what is the minimum commitment?
• What results would make you recommend stopping the test?
• Which of the brands you cite have live campaigns, and can you share results they have agreed to make public?

A good partner answers these specifically. A vague answer about 'AI visibility' is a reason to run a small self-serve test first and learn the channel yourself.

Where your collection pages come in

Conversational discovery tends to start broad: 'something to wear to a sangeet', 'a skincare set for dry skin', 'a minimal watch for office'. The natural landing page for a broad question is a collection, not a single product. That makes the order of products on your collection pages part of your ad performance, whether you think of it that way or not. A shopper who lands on 'Sangeet outfits' and sees out-of-stock items and last year's slow sellers in the first row is likely to leave, however good the ad was.

SellRank exists for that first row. It ranks any Shopify collection by measured performance, such as conversion rate, revenue, add-to-carts or profit, lets you compose a page with percentage slots and conditions like in stock only, and syncs the order into Shopify's own sort with no theme edits. Its built-in pixel counts views and add-to-carts per product per day, aggregates only, so ranking by engagement works from day one. On paid plans it keeps version history and per-version performance, so you can see what changed when a campaign started sending traffic.

The honest summary

ChatGPT ads in India are real, live and, at the reported minimum, cheap enough to test. Razorpay's 5 October partnership makes them easier for brands that want someone else to run them. But the channel is new, formats and self-serve details are still settling, and published results from Indian brands do not yet exist in the sources we checked. The work that will decide whether it pays for your brand is mostly work on your own store: accurate product data, honest prices, fast pages and collection pages that lead with the right products.

If you want help getting your Shopify catalogue and collection pages ready for AI discovery, paid or organic, tell us about your store on our contact page. If you want your collection pages to lead with what converts now, try SellRank on its product page; the free plan covers one collection with the full editor and one live sync.