Retail media did not win because it is a better advertising channel; it won because it is the first measurement system Indian consumer-goods brands have ever had that actually sees the sale.
Consider where an Indian fast-moving consumer goods (FMCG) marketing head sits in 2026. Advertising money flows to Amazon Ads, Flipkart Ads, the quick commerce networks built by Blinkit, Zepto and Swiggy Instamart, and the vertical platforms in beauty and fashion. Each of these will report, inside its own dashboard, how many units were sold to shoppers who saw the sponsored placement. That sentence would have been science fiction to the same marketing head in 2015 — not because the analytics were unavailable, but because the transaction itself was invisible to the brand. India compressed roughly a decade of Western retail media development into the years between 2020 and 2025, and it did so on a structure of its own: horizontal marketplaces, category verticals and ten-minute delivery apps rather than the grocery chains that anchor the Western version.
To understand why this matters, you have to remember what the problem actually was. Across a decade of deployments in Indian consumer goods, the pattern repeats: the brand has enormous equity, tens of millions of daily consumers, and almost no first-party record of any of them buying anything. The reason is structural, not technological. India’s consumer-goods volume moves through an indirect chain — distributor to wholesaler to kirana — and the manufacturer never touches the end transaction. The mid-2010s channel arithmetic looked roughly like this, and it framed every analytics conversation of the period:
| Channel (India FMCG, mid-2010s, illustrative estimates) | Share of volume | Consumer transaction data available to the brand |
|---|---|---|
| Kirana / general trade | ~75–80% | None |
| Modern trade | ~15–20% | Category-level at best, never consumer-level |
| Owned channels | Under 1% | Full — but negligible volume |
| E-commerce | Under 2% (2015) | Partial |
Every serious analytics programme of that era was, honestly described, an attempt to approximate a fact nobody could observe. Marketing mix modelling (MMM) decomposed aggregate sales into channel contributions precisely because it needed no consumer-level data — an elegant statistical workaround for a missing measurement, not a measurement. Social listening was sold as a consumer-signal proxy: if you cannot see transactions, at least see what people say about them. Trade analytics command centres dashboarded distributor and field-sales data because that was the data a brand could legally and practically get — a trailing indicator of what moved through the pipe, not what consumers chose. Pack-level registration pilots tried to make the product itself the data-capture instrument. All of it was intelligent. All of it was proxy work. The stacks of that period — statistical modelling suites, campaign platforms, integration middleware — were being asked to manufacture attribution out of data that structurally did not contain it.
Here is the dual lens we apply at ThirdVoice. The brand argument for retail media is often made suspiciously — that it is a bottom-funnel tax, a toll the marketplaces extract for shelf position, money that builds no equity. There is truth in the complaint about pricing power. But the evidence argument is decisive on a different axis: retail media is the only environment in which the ad exposure and the purchase live in the same data system. The attribution chain — shopper sees the sponsored listing, clicks, buys, and the platform reports ad-attributed units — requires no identity resolution project, no data integration programme, no modelled inference. What the 2015 stacks tried to assemble from proxies, the platform produces natively, because advertising and transaction were vertically integrated into one environment. The fix for the indirect-market attribution problem turned out not to be better analytics middleware. It was a structural change in where the advertising happened.
This is, in our judgement, the most significant structural shift in how Indian consumer-goods brands measure marketing effectiveness in the past decade — more consequential than any single platform purchase. And it carries an uncomfortable corollary for anyone who spent the 2010s buying attribution technology: the problem was never solvable at the brand’s end of the pipe. A conglomerate whose consumers transact anonymously in cash at a kirana could have deployed every modelling suite on the market and still never closed the loop, because the loop’s other end — the transaction — belonged to someone else.
Two cautions before this becomes triumphalism. First, closed-loop is not the same as incremental. The platform that sells you the media also grades the homework; ad-attributed units include buyers who would have purchased anyway, and the sponsored slot at the top of a search for your own brand name harvests demand you already own. The disciplined operators now insist on incrementality testing and treat platform-reported return as a ceiling, not a truth. Second, retail media measures the channel it lives in. The kirana, still the majority of Indian FMCG volume, remains as dark as ever; the Goods and Services Tax (GST) trail and business-to-business trade platforms have illuminated the retailer layer, not the consumer one. MMM, ironically, is back in fashion globally precisely to referee across channels that retail media cannot see. The closed loop is real, but it closes over a minority of your volume — a fast-growing, urban, disproportionately valuable minority.
What should a senior leader do differently on Monday? Stop budgeting retail media as a sales-team discretionary line and start treating it as measurement infrastructure: the one channel where a rupee of spend produces an observable, auditable sales outcome. Use that observability aggressively — demand test-and-control incrementality studies from every network you spend with, and use the calibrated results to discipline the claims of every other channel on your plan. Fragmentation across eight-plus Indian networks is now an operating problem, so assign ownership: one accountable owner for retail media spend, creative and measurement across platforms, not a per-marketplace scatter. And retire, formally, any attribution programme whose honest description is “a proxy for the transaction we cannot see.” For the first time, in one channel, you can see it. Build around that.