Retail Media in India: What You Are Actually Buying

Open a marketplace app and search for anything. Most of the first screen is paid. Not the platform advertising itself — the brands it stocks, bidding to sit above each other.

That is retail media — also called commerce media, or marketplace advertising when the surface is a marketplace. This page covers what you are buying, who sells it in India, and whether it earns a line in your plan. It sits inside our performance marketing guide.

One thing to clear up before anything else. Retail media is not retail marketing. Retail marketing is what a retailer does to sell its own stock, across its store and its own website. Retail media is what a brand buys on a retailer's platform. Different buyer, different budget line.

What a retail media network actually is

A retail media network is the advertising business a retailer runs on its own property. It sells three things it owns and you do not: shopper traffic that arrives with buying intent, purchase data on real people, and the ability to report a sale that followed an ad.

Retail media networks package that into three kinds of placement:

• On-site search and listing placements. Your product appears above or inside the results for a query, usually on a cost-per-click auction. Amazon Ads says of its sponsored ads: "Pay only when the ad is clicked" (Amazon Ads, Getting started with sponsored ads, accessed 15 October 2026).
• On-site display and brand space. Banners, category takeovers, video and a brand store page. Flipkart's ad formats include carousel ads, interactive shoppable videos, gamified ad experiences and 3D product explainers, per Vijay Iyer, VP and GM of Flipkart Ads (exchange4media, 26 May 2025).
• Off-site extensions. The retailer's shopper data used to target ads on the open web or on other apps. That is ordinary programmatic buying with a retailer's audience attached — the mechanics are in our programmatic advertising in India post.

The seller writes the scorecard.

India's retail media surfaces, by type

Four surface types matter, and they behave differently. Pick by where your category is shopped, not by platform size.

Sources for two of those cells: Myntra Ads cited video ads, polls and rich content formats (exchange4media, 29 January 2024); Zepto launched Jarvis as its in-house ad platform (Storyboard18, 25 November 2024). Elsewhere we verified only that the platform sells placements — ad menus are gated and change, so read the platform's own ads page.

Why every Indian app became an ad business

Not fashion. Margin. Retail margins are thin and advertising margins are not, so the ad line fixes the retail line.

Swiggy said as much about Instamart: "Growth in advertising has been a key driver for take-rate expansion." Its reported Instamart take rate moved from 12.2% in Q1 FY24 to 15.2% in Q2 FY25, with a steady state of 20–22% expected (Storyboard18, 4 December 2024, on Swiggy's Q2 FY25 results). Read that sentence as an advertiser: the ad budget is part of what the platform takes from your order.

Two structural assets make it sellable:

1. First-party purchase data. The platform knows what people bought, how often and in which pincode — not a modelled guess. Flipkart's Shopper Audience Platform offers over 1,000 audience segments drawn from 500 million-plus registered users, per its ads VP in the exchange4media interview above. Why that data is valuable at all is covered in our first-party data post.
2. Closed-loop attribution. The ad and the purchase happen in the same app, so the platform can join them without cookies. Flipkart reports impressions, clicks, add-to-cart rate and units sold in one view. That join is the real product being sold.

Demand followed. Blinkit reported 557 advertisers in Q3 FY24, up from 242 a year earlier, with ad revenue growing faster than its order value — though it did not disclose the rupee figure (Storyboard18, undated page, accessed 15 October 2026).

The honest caveat: the platform marks its own homework

Closed-loop attribution is genuinely better than what the open web offers. It is also graded by the party you are paying.

The platform sets the attribution window, decides what counts as an ad-driven sale, and chooses which numbers you see. A high return on ad spend inside a marketplace dashboard can simply mean the ad was shown to people who were already walking to your shelf. Nothing is being faked — the measurement just cannot tell demand it created from demand it caught.

Three things reduce that risk:

• Hold the platform's number next to your own. Compare reported ad-driven units against total units for the same SKU and week. Our marketing attribution post covers reconciling two sources that disagree.
• Test, do not argue. Switch a campaign off in some pincodes and keep it on in others, then look at total sales rather than attributed sales. That is incrementality testing, and it is the only answer to "would we have sold it anyway".
• Get the definition in writing. Attribution window, view-through treatment, whether branded search counts. If nobody can state it, assume the generous version.

What you control, and what the platform controls

Most disappointment comes from assuming it works like a self-serve ad account you own. It does not.

That last column is not theoretical. Zepto was reported to have auto-activated campaigns for brand partners that had none running, with a note saying "we've proactively activated auto campaigns for your account"; sellers said default budgets, bidding and targeting were not disclosed, and that opting out meant justifying it to an account manager (Storyboard18, undated page, accessed 15 October 2026).

So your real control is the wallet, not the campaign. Fund it deliberately, check it weekly, and keep the power to switch spend off with one person who watches margin — the discipline in media buying.

When retail media earns budget — and when it is a tax

It earns the budget when four things are true at once:

• You are stocked and in stock. An ad on an out-of-stock or unavailable-in-pincode SKU buys nothing. Check availability before bids.
• The category is genuinely shopped there. Staples and top-ups on quick commerce, model-number searches on horizontal marketplaces, browse-led fashion and beauty on vertical ones.
• The listing already converts. Ads multiply a conversion rate; they do not create one. Fix titles, images, reviews and price first — that work sits in ecommerce marketing.
• The margin survives the stack. Commission plus fulfilment plus returns plus ad cost, before you call it profitable.

It is a tax you pay to a marketplace when any of these are true:

• You are bidding on your own brand name only because a competitor appeared there, and spend keeps rising with no test behind it.
• The target is set by the platform's dashboard rather than your contribution margin.
• Spend auto-scales during a sale event and nobody reconciles it afterwards.
• The platform owns every customer you acquire, so none of the spend compounds. A brand with no direct channel has no fallback — the reason D2C marketing strategy exists alongside marketplace selling.

Our own split between marketplace spend and brand-owned channels

How we decide it (Digital Hangover's own view, no client figures attached): we do not start from a percentage. We start by asking what each rupee still does if the platform changes its rules next quarter, then set a ceiling rather than a target.

Three practical rules we apply:

1. Cap marketplace ad spend near the share of revenue that platform already produces. If a marketplace is a fifth of revenue, it should not be taking most of the media budget. Growth beyond that cap has to be proven with a holdout test, not a dashboard.
2. Ring-fence a fixed slice for brand-owned demand. Search, your own site, email and owned audiences. This is the spend that still works if an account gets throttled or a fee schedule changes.
3. Judge the two on different questions. Marketplace ads answer "did we win the shelf this week". Brand-owned spend answers "is anyone asking for us by name". Averaging them into one blended ROAS hides both.

Working through the maths (hypothetical numbers, to show the method): take a ₹1,000 order with a 45% gross margin, so ₹450 to play with. Platform commission and fulfilment at 20% takes ₹200. Returns at 8% of orders cost roughly ₹35 per order spread across the base. That leaves about ₹215. A ₹60 ad cost per order leaves ₹155 — workable. At ₹200 ad cost per order you are buying revenue with your own margin. Run that single line before you accept any ROAS target.

Use your own commission, fulfilment and return rates. Rupees left per order is the number, not the dashboard multiple.

Building a retail media plan that survives a rule change

Retail media is not optional for a brand sold on these platforms — if you do not bid, a competitor sits above your listing. It is also the one channel where your landlord is also your media seller. Both can be true.

So treat it as a bought shelf with a rate attached, not a growth engine. Google Shopping ads and your own site reach the same shopper outside the platform's auction. Platform-level detail sits in Amazon ads in India and Amazon marketing.

Do that, and in two quarters you will know something most brands on these platforms cannot answer: which part of your marketplace revenue you are buying, and which part you actually own.