Retargeting Marketplace Shoppers When You Do Not Own Their Data

A founder we work with put it bluntly last quarter. We spend lakhs sending traffic to our Amazon listings, half of them bounce without buying, and we cannot do a single thing to chase them. He was right, and that frustration is the whole problem with marketplace retargeting. On your own website you drop a pixel, build a list, and follow the shopper around the internet for weeks. On a marketplace you cannot. The buyer is Amazon’s customer, not yours. You never see the email, you never see the phone number, and you are not allowed to build your own remarketing list from their behaviour. So the instinct is to give up on retargeting entirely. That instinct is wrong. You can retarget marketplace shoppers. You just have to do it with the platform’s audience pools instead of your own data, and you have to be deliberate about which pools you reach for.

Why first-party retargeting is off the table here

Start with the constraint, because everything else follows from it. When someone buys your product on Amazon or Flipkart, the marketplace owns the transaction and the customer relationship. You get an order to fulfil, sometimes a masked contact detail, and almost nothing you can legally turn into a marketing list. There is no pixel of yours firing on the product page. There is no cart-abandon event landing in your own systems. The shopper who viewed your listing three times and left is invisible to you at the individual level.

This is not a bug you can engineer around. It is the deal you accepted when you chose to sell on someone else’s storefront. The same opacity that blocks retargeting also makes measurement harder, which is why we treat marketplace analytics as a separate discipline from D2C analytics. If you have ever tried to run real cohort analysis without first-party data, you already know how thin the visibility is. Retargeting lives under the same ceiling.

The audiences you are actually allowed to use

Here is the part most brands miss. The marketplace will not hand you the data, but it will rent you access to the audiences built from that data. Amazon in particular has assembled enormous behavioural pools, and through its demand-side platform it lets you target them with display and video. You are not retargeting your shoppers. You are retargeting Amazon’s segments that contain your shoppers. The distinction matters because it changes what you can and cannot ask for.

The pools worth knowing, roughly in order of intent:

• Product viewers. Shoppers who looked at your detail page and did not buy. This is the closest thing to classic cart-abandon retargeting that a marketplace offers, and it is usually the highest-return audience you can buy.
• Past purchasers. People who already bought from you. In repeat-friendly categories this is gold, because reactivating a known buyer is cheaper than winning a stranger.
• Category browsers. Shoppers active in your category who have not yet found you. Lower intent than your own viewers, but warmer than cold prospecting.
• Competitor viewers. Audiences who looked at similar products. Useful for conquesting, expensive to convert, handle with care.
• Lookalike and in-market segments. Modelled audiences that resemble your buyers or are signalling purchase intent in your space.

Notice that all of these are platform-defined. You never see who is in them. You choose a pool, you choose creative, and the marketplace matches. That is retargeting without ownership, and it is the only kind available to you here.

You are not chasing your customer across the web. You are renting the marketplace’s memory of them, one audience pool at a time.

Amazon DSP is the main door, not the only one

For most Indian brands serious about this, the practical vehicle is Amazon’s demand-side platform. It is the sanctioned way to act on those audience pools at scale, and the retargeting use case is where it earns its keep. But DSP is not a starter tool, and reaching for it too early wastes money. The viewer and purchaser pools have to be large enough to be worth addressing, which means you need real traffic flowing through your listings first. We have written the full threshold case in when Amazon DSP is actually worth it, and the short version is simple. If nobody has viewed your listings yet, there is nobody to retarget, and DSP has nothing to work with.

Below the DSP threshold you are not without options, you just have blunter ones. Sponsored Brands and Sponsored Display offer lighter retargeting-style placements that recirculate some shopper attention without the overhead. Sponsored Display in particular has views-remarketing and audience targeting that smaller brands can switch on long before DSP makes sense. It is coarser, but it touches the same logic of re-engaging shoppers the platform already knows.

How to read the returns honestly

Retargeting on a marketplace breaks last-click reporting in a specific way, and if you do not adjust for it you will switch off the thing that is working. A display impression that nudges a shopper back to search for your brand often gets no credit, because the conversion is attributed to the branded search click that followed. The retargeting did the work. The search bar took the medal.

So judge these audiences on blended performance, not isolated last-click return. Watch what happens to branded search volume, to repeat-purchase rate, and to total category sales when you turn a retargeting pool on and off. This is also why the economics only make sense once you understand value over the full relationship, not the single order. Retargeting past purchasers is rational precisely because a marketplace buyer is worth more than one transaction, the argument we lay out in estimating customer LTV on marketplaces. If you price a repeat buyer at their first order alone, you will underbid every reactivation audience and lose them to competitors who did the maths properly.

A deliberate sequence, not a switch

The mistake we see most is treating marketplace retargeting as one button labelled on. It is a sequence, and the order protects your budget. The way we stage it:

1. Build the pool before you spend on it. Get enough listing traffic and purchase volume that the viewer and buyer audiences are large enough to be addressable. Until then, put the money into search.
2. Start with the highest-intent pool. Product viewers first, then past purchasers in repeat-friendly lines. Prove the return before widening.
3. Widen outward only as efficiency holds. Category browsers and in-market segments next, conquesting last and cautiously.
4. Read it blended from day one. Decide on total business impact, branded search lift, and repeat rate, never on a tidy last-click number.

Do it in that order and each pool earns the right to the next. Do it backwards, spending on broad cold audiences before your own viewers exist, and you will conclude marketplace retargeting does not work, when really you just used it inside out.

What changed recently

The reason this matters more in 2026 than it did two years ago is that the audiences you are renting have become the marketplace’s most valuable product. Retail media is now the fastest-growing slice of Indian adex. Amazon and Flipkart together booked roughly Rs 14,652 crore in ad revenue in FY2025, with Amazon India’s ad sales up 25 percent to Rs 8,342 crore, per Exchange4media. When advertising is already near a quarter of a platform’s operating base, the audience pools get richer, the tooling gets better, and the pressure on you to use them well goes up, not down.

Amazon has also been rebuilding the machinery you actually touch. Through 2025 it folded display retargeting into a unified Campaign Manager workspace and extended in-market audience targeting across markets including India, with its DSP moving toward general availability and a phased migration into 2026, according to Amazon Ads. The practical effect for a mid-sized brand is that the gap between coarse Sponsored Display remarketing and full DSP is narrowing, so the viewer and purchaser pools are reachable earlier in your journey than they used to be.

The bigger shift is where the next pool is being built. Quick commerce has turned into a serious ad surface of its own. Blinkit, Zepto and Instamart are projected to draw around Rs 4,900 crore in ad revenue in 2026, and industry operators told Storyboard18 that between 10 and 25 percent of digital performance budgets are already shifting onto these platforms. The retargeting logic carries straight across. You still do not own the buyer, the platform still rents you its memory of them, and if you sell in impulse categories the same sequencing discipline applies on a 10-minute storefront. If you are weighing where to point first, our note on marketing a brand on quick commerce in India sits right next to this one.

Where this sits in the wider plan

Retargeting is one instrument inside the channel, never the whole performance programme. It only pays back when your search foundation is already efficient, and it has to be coordinated against everything else competing for the same rupees, which is the discipline we describe in running performance across marketplaces on one budget. That coordination is the core of our Performance Marketing & Ads practice, and the display side leans hard on our Creative & Content Studio, because a retargeting banner with weak creative converts nobody no matter how warm the audience. The data limits sit alongside the measurement work in our Analytics & Marketplace Intelligence thinking, since you cannot judge a pool you cannot directly see without a deliberate read of the signals you do get.

The summary is plain. You do not own the marketplace shopper, and you never will. But the platform remembers them, and it will rent you that memory in defined pools. Treat those pools the way an operator treats any borrowed asset. Know exactly what each one is, reach for the warmest first, pay for them based on the full relationship and not one order, and measure them on what they do to the whole business. Retargeting without first-party data is not a workaround. On a marketplace, it is the only game, and it is a good one once you play it in the right order.