ZANE™. EST2016.
waiting for you Book a meeting →

Scaling From One Crore to Ten on Marketplaces Without Breaking Ops

Getting a marketplace brand to one crore in annual revenue is a demand problem. You find a category with room, get the listings right, fund ads behind the SKUs that convert, and ride the platform’s traffic. Most of the work sits at the top of the funnel, and the constraint that holds you back is almost always whether enough people are finding and buying the product. So that is where founders point their attention, their budget, and their hiring. It works. It gets them to a crore.

Then they try to do it ten times over, with the same playbook, and the brand starts to crack in places that have nothing to do with demand. Stockouts on the hero SKU during the exact week ads are scaled. Cash locked in inventory the brand cannot reorder. A returns pile that quietly eats the margin the topline was celebrating. The demand was there. The brand could not carry it. That is the whole story of scaling on marketplaces in India, and the brands that survive it are the ones who planned for the constraint they had not hit yet.

The bottleneck moves, and most founders chase the old one

At one crore your binding constraint is demand. At ten crore it is almost never demand. It is some combination of inventory, cash, and process. The danger is not that the bottleneck is hard to fix. It is that it moves silently, and a founder trained by the first crore keeps optimising the thing that is no longer the problem.

You see it in the ad account. The brand that cracked acquisition keeps pushing spend because spending is the lever they know, and the orders arrive faster than the supply chain can serve them. Now you are paying to acquire demand you cannot fulfil, your account health drops on late shipments, and the platform throttles the listings you just paid to promote. The ads were never the problem. The ceiling was downstream the whole time.

Scaling does not break where you are looking. It breaks where you stopped looking once the first crore came in.

So the first discipline of scaling is unglamorous. Before you push more demand, find out what would fail if the demand doubled tomorrow. Usually it is not the funnel. It is the warehouse, the reorder cycle, or the bank balance.

Operations is the first wall you hit

The operational load of a marketplace brand does not rise in line with revenue. It rises faster. Ten times the orders is more than ten times the support tickets, returns, reconciliation lines, and SKU-location combinations to keep in stock. Process that ran on a founder’s memory and a shared spreadsheet at one crore becomes the single point of failure at five.

The brands that scale cleanly tend to have built the boring parts before they needed them. Inbound and labelling that does not depend on one person. A returns process that inspects, grades, and restocks instead of writing everything off. Reconciliation that catches the fees and deductions the platform quietly applies. None of this is exciting, and all of it is the difference between growth that compounds and growth that collapses under its own order volume. We made the full argument for front-loading this in the operations setup checklist before you list a single SKU, and at scale the cost of having skipped it is simply larger.

The signal that operations has become your bottleneck is specific. Your account health metrics start sliding even though nothing changed in the product. Late dispatch rate creeps up. Cancellation rate ticks. Negative feedback clusters around delivery, not the item. Those are not customer-service problems. They are capacity problems wearing a customer-service mask, and no amount of ad spend fixes them.

Capital is the constraint nobody plans for

This is the one that catches good brands by surprise, because it is invisible on the topline. Marketplace growth eats cash. You pay for inventory now and collect from the platform later, on a settlement cycle that can run a week or more. The faster you grow, the bigger that gap, because every rupee of new revenue requires inventory funded before the cash from the last batch has landed. Profitable brands run out of money mid-scale all the time, and the P&L gives no warning because the problem is not profit, it is timing.

Working capital, not demand, is what actually rations how fast you can grow once the funnel works. A brand that can fund three inventory cycles can scale three times faster than an identical brand that can fund one, regardless of how good either one’s ads are. That single fact reorders the priority list for most founders, and we devoted a whole piece to why in working capital is the real constraint on marketplace growth. The math is unforgiving and it does not care how strong your conversion rate is.

The practical moves at this stage are about widening the gap you can survive.

  • Forecast inventory against the growth you are funding, not the growth you hope for. Order to a demand plan you can actually pay for, because an out-of-stock hero SKU costs you rank that takes weeks to rebuild.
  • Negotiate supplier terms before you need them. Thirty days of credit from a manufacturer is cheaper working capital than anything a lender will offer, and it scales with the relationship.
  • Treat the platform settlement cycle as a financing cost. Reconcile it, claim back every wrongful deduction, and know to the day when cash lands so you can time reorders against it. Faster settlement tiers are a real working-capital lever, and they are earned through fulfilment reliability, not negotiated.
  • Hold a cash buffer sized to one full inventory cycle. The brands that die mid-scale are usually the ones who ran the buffer to zero to fund one more batch.

The long tail gets heavier as you grow

At one crore, a sprawling catalogue is survivable. A few slow SKUs hide inside the topline and nobody notices the cash they are holding hostage. At ten crore the same long tail is a serious drag, because every dead SKU is inventory you funded, warehouse space you rented, and reconciliation lines you process, all returning nothing. The catalogue that felt like optionality at the start becomes ballast.

Scaling is partly an act of subtraction. The brands that grow cleanly are usually narrower than they were, concentrating cash and attention on the SKUs that actually compound and cutting the ones that only consume. That decision gets harder emotionally as the catalogue grows, which is exactly why it needs a method rather than a feeling. We laid one out in killing the long tail that is bleeding you, and at scale the discipline pays back in freed capital faster than almost anything else you can do.

What changed recently, and what it means for your scaling math

Two shifts in the last year move the numbers in this article, and both cut in the brand’s favour if you are paying attention.

The first is fees. Amazon India has been cutting seller economics aggressively. Effective March 2026, it expanded zero referral fees to over 12.5 crore products priced under one thousand rupees across more than 1,800 categories, alongside lower Easy Ship and weight handling charges, and says the combined changes can save sellers up to 70 percent in total selling fees on qualifying items, per Amazon India. That is not a reason to relax. It is a reason to rerun the contribution margin on every sub-thousand-rupee SKU, because a SKU that was marginal at the old take rate may now be worth funding, and the brands that re-cut the math first will redeploy that freed margin into rank before slower competitors notice.

The second is that the demand surface itself is shifting toward quick commerce, and the platform economics there are tightening in the other direction. Quick-commerce players have spent the last two years layering on handling, platform, and surge fees to fix their own unit economics, as Storyboard18 documented, and the leaders are now turning structurally profitable on the back of it. Blinkit posted its first positive adjusted quarterly EBITDA, around four crore rupees, in the December 2025 quarter while running past two thousand dark stores on an increasingly inventory-led model, per Storyboard18. For a brand, an inventory-led quick-commerce platform means the platform, not you, increasingly decides what gets stocked in each dark store, which makes availability and trade terms the new battleground. We work through what that does to your margin in quick-commerce unit economics after platform fees. The headline for scaling is simple. The channel that grows your demand fastest is also the one that compresses your control hardest, so the operations and capital discipline this whole piece argues for matters more there, not less.

Plan the constraint, do not just react to it

The mistake is to scale until something breaks, fix it, and scale again until the next thing breaks. That works, badly, and it costs you a cracked listing rank or a cash crunch each time. The better operators do something different. They look one stage ahead and ask what the binding constraint will be at the revenue they are aiming for, then build for that constraint before they arrive at it.

That is why the order of operations matters more than the speed. Before you double ad spend, confirm the supply chain can serve the orders. Before you broaden the catalogue, confirm the cash can fund it. Before you enter a second marketplace, confirm operations can carry two account structures without the first one slipping. Sequencing the constraints is the actual skill, and it is the spine of a twelve-month marketplace growth roadmap that survives contact with reality. A roadmap that plans demand without planning the operations and capital to carry it is a wish, not a plan.

What changes in how you spend your own time

The founder who scales well usually stops being the best media buyer in the building and becomes the person who watches where the next wall is. Less time in the ad account, more time on the reorder calendar, the cash position, and the account-health dashboard. The leverage moves from acquisition to throughput, and so should your attention. The demand that got you to one crore is rarely what stops you from reaching ten. The thing carrying that demand is.

The short version

One crore is a demand problem and ten crore is an operations and capital problem, and the failure mode is bringing the first playbook to the second fight. The bottleneck moves from the funnel to the warehouse to the bank, quietly, while you are still optimising acquisition. Fee cuts on platforms like Amazon and the rise of inventory-led quick commerce change the numbers but not the lesson. Plan the constraint you will hit next, build for it before you arrive, and sequence growth around what the brand can carry rather than what the funnel can produce.

This is the work our D2C & Marketplace Strategy Consulting exists to do, mapping which constraint binds next and building for it ahead of time. Our Marketplace Performance teams keep the demand engine running while our Analytics & Reporting work makes the moving bottleneck visible before it breaks something. Demand gets you started. Operations and cash decide how far you go.

The Subscribe-and-Save Lever Most FMCG Brands Ignore on Amazon

Most FMCG brands on Amazon India treat every order like a fresh acquisition. They pay to win the click, pay to win the buy box, and then watch the customer disappear into the marketplace fog. For a coffee brand, a protein brand, a diaper brand, a vitamins brand, this is quietly absurd. The product is bought again and again on a clock. Yet the brand reacquires the same buyer every cycle as if they were a stranger. Subscribe and Save fixes this, and it sits unused on most listings.

Why subscriptions matter more than the discount they cost

Subscribe and Save asks you to give up a small percentage in exchange for a standing order. On the surface that reads like a margin leak. Look at it as an operator and it reads like the cheapest repeat-purchase mechanism Amazon will ever hand you. You are converting a one-time buyer into a recurring one without paying ad spend on the second, third, and tenth order.

That last point has stopped being theoretical and become urgent. On quick commerce, the channel everyone is chasing, FMCG margins have fallen three to five percentage points in recent months as brands bid for visibility, with peak-slot ad costs nearly doubling and total category ad spend crossing five thousand crore rupees in FY25, according to The Economic Times. When the cost of being seen keeps climbing, a customer who reorders on autopilot without you paying for the impression is not a nice-to-have. It is the margin you have left.

The math compounds. A single conversion to a subscription is not one extra sale. It is a chain of sales that lands on a schedule you can see in advance. That changes how you value the first order entirely. We dig into this in our piece on customer LTV on marketplaces, but the short version is that a subscribed customer is worth a multiple of a one-off, and the discount you concede is small against that.

The forecasting payoff brands underrate

The revenue story gets the attention. The forecasting story is the one that actually saves operations teams from chaos. Demand for consumables on Amazon India is spiky. Festival pushes, a viral review, a competitor stockout, all of it whips your demand curve around. Subscriptions are the part of your volume that does not whip. They arrive on a known cadence.

A subscription base is the closest thing a marketplace brand has to a salaried customer. Everything else is freelance.

When a meaningful slice of your monthly volume is on standing order, your inventory forecasting stops being a guessing game on that slice. You know roughly how many units ship on the renewal dates. You buy raw material against it. You hold less safety stock because the predictable portion does not need a buffer. Stockouts on subscribed SKUs are far more damaging than on regular ones, so this predictability is not a luxury. It protects the very base you worked to build.

Who should be using it and is not

The list of brands that should run Subscribe and Save and do not is long. If your product is consumed and replaced, you qualify. A few categories where we see the gap most often:

  • Grocery and pantry staples like coffee, tea, oils, spices, and dry snacks bought on a monthly rhythm.
  • Health and supplements where a bottle lasts thirty days and adherence depends on never running out.
  • Baby and personal care such as diapers, wipes, and skincare with a fixed burn rate.
  • Pet food, which is one of the most reliably recurring purchases in any basket.
  • Home and cleaning consumables that customers hate having to remember to reorder.

If you are a new consumable brand still deciding where to plant your flag, the subscription motion should be designed in from day one. The fast channel and the marketplace are not rivals here. Lead with quick commerce for trial and impulse, then use Amazon subscriptions as the retention layer underneath it, and decide the split with the kind of thinking in our note on the marketplace versus D2C margin tradeoff.

How to actually make it work, not just enable it

Toggling Subscribe and Save on is not a strategy. We have seen brands enable it, never mention it again, and wonder why almost nobody subscribes. The lever only pays off when you treat it as a managed program.

Make the offer visible

The subscription option lives on the listing, but most shoppers default to a one-time purchase out of habit. Your A+ content, your images, and your bullets should make the standing-order value obvious. Show the price after the subscription discount. Spell out that they can pause or cancel anytime, because the fear of being locked in kills more sign-ups than the price does.

Tune the discount tier

You control how much you concede. Set it too low and there is no reason to commit. Set it too high and you erode margin on a customer who would have repurchased anyway. The right number depends on your repeat-purchase rate and your contribution margin per unit. This is exactly the kind of decision that belongs inside a real Marketplace Account Management motion rather than a one-time setup.

Match the pack to the cadence

Quick commerce has already pushed brands to re-engineer pack architecture, splitting hero SKUs into smaller, channel-specific sizes built for instant baskets, as Business Standard reported. Apply the same discipline to subscriptions. The pack you put on a standing order should map to a real thirty-day burn rate, not your default retail size. A bottle that runs out at day twenty breaks the renewal rhythm. One that lasts forty-five days trains the customer to skip.

Protect the subscribed SKU

Never let a subscribed SKU go out of stock. A missed renewal is not a delayed sale, it is a cancelled relationship. Customers who get a failed delivery often drop the subscription entirely. The predictability you gained becomes the liability you ignored.

Reading the base without first-party data

Amazon does not hand you a clean customer database, so brands assume they cannot measure retention. You can, just indirectly. Subscription enrollment and renewal counts are a signal you do own, and they are the cleanest retention proxy the marketplace gives you. Pair that with the approach in our guide to cohort analysis without first-party data and you can watch how each month of new subscribers holds up over time. That tells you whether your product genuinely retains or whether the first reorder is where people quietly leave.

This matters because a healthy subscription base is also a product-quality verdict. If subscribers churn fast, the discount is not the problem. The product, the pack size, or the price point is. Subscriptions surface that truth earlier than a pile of one-off orders ever would.

What changed recently

The subscription case has only gotten stronger as the channel mix shifts under FMCG. A few developments worth holding in view:

  • Amazon is building its own instant layer. Amazon Now, its quick-commerce service, is expanding to 100 cities backed by more than a thousand micro-fulfilment centres and a roughly 2,800 crore rupee investment, per Inc42. As instant and scheduled delivery converge inside one ecosystem, a brand with a managed subscription base is positioned to ride both rather than pick one.
  • Reorder cycles are compressing. Quick commerce has pulled FMCG repeat-purchase intervals down to as little as five to seven days for staples that used to reorder every twelve to fifteen, Business Standard notes. Faster natural cadence makes a standing order easier to justify to the customer and more valuable to you.
  • Visibility is no longer cheap. With ad costs eroding channel margins, the orders you do not pay to win are worth more than they were a year ago. A subscription is the clearest version of an order you do not re-buy every cycle.

The operator takeaway

Subscribe and Save is not a feature you switch on and forget. It is a lever you manage. Done well, it turns volatile demand into a base you can forecast, lowers your blended acquisition cost, and gives you an early read on whether customers actually keep coming back. For a consumable brand, ignoring it is leaving the most defensible part of your revenue unbuilt.

If you sell something people finish and rebuy, the question is not whether to run subscriptions. It is how much of your category you are willing to let competitors lock in first. Treat it as a managed program inside your Marketplace Account Management and your Marketplace Advertising strategy, and the discount stops looking like a cost and starts looking like the price of a customer who pays you on a schedule.

A 12-Month Marketplace Growth Roadmap That Survives Contact With Reality

Most twelve-month marketplace plans are works of fiction. They open with a modest month one, then bend upward into a clean hockey stick by quarter four, as if demand were the only constraint that mattered. We have watched enough of these plans meet reality to know what happens next. Demand was never the bottleneck. Inventory was. Cash was. The two-person ops team was. The plan assumed a frictionless climb and the business hit a wall it had drawn on the chart as open road.

A roadmap that survives contact with reality looks different. It does not promise a number. It sequences a set of phases, and it gates each phase behind a capacity threshold you have to actually clear before you are allowed to push the next lever. Launch, stabilize, scale. In that order, and never out of it. Here is how we build one that holds.

Why the hockey stick lies to you

The hockey stick is seductive because it is easy to draw and pleasant to present. It also encodes a dangerous assumption: that growth is a function of ambition rather than capacity. On marketplaces, that assumption gets punished fast. You scale ad spend before your listings convert, and you pay to send traffic to pages that do not earn it. You chase a festival spike before your fulfilment can absorb it, and you take a hit on your account health that costs you the next quarter.

The honest version of a roadmap treats every growth lever as something that loads weight onto a system. Before you pull the lever, you ask whether the system can carry the weight. If the answer is no, the lever waits. This is the discipline that separates a plan from a wish, and it is the same operator instinct behind the operator-led agency model and why doers beat decks. People who have run the ops do not draw lines they cannot defend.

A roadmap is not a forecast of how fast you want to grow. It is an honest map of how much weight each part of your business can carry before it breaks.

Phase one: launch, months one to three

The first quarter is not about revenue. It is about proving the unit works. You are establishing that a single product, on a single marketplace, can convert paid and organic traffic into orders at a margin you can defend. Resist the urge to be everywhere at once. Breadth in the launch phase is how brands spread themselves thin and learn nothing clearly.

Pick the one marketplace and the few hero products where you have the strongest right to win. If you are unsure how to choose, our marketplace prioritization framework for resource-strapped brands exists precisely for this decision. The launch phase is also where readiness is non-negotiable. Going live with half the inputs in place is the most common way to poison the next nine months.

  • One marketplace, a tight hero range, and listings that are fully built and out of category review.
  • Brand registry approved and inventory physically received, not in transit, before the button gets pushed.
  • A narrow ad plan to gather conversion data, not to dominate search on day one.
  • Account health watched daily from the first order, because the early signals compound.

Work through a structured brand launch readiness checklist for Indian marketplaces before you go live. The gate out of phase one is simple: a proven conversion rate, clean account health, and a margin you can live with at volume. If you cannot show those three, you do not graduate to stabilize. You fix launch first.

Phase two: stabilize, months four to seven

Stabilize is the phase everyone wants to skip, and skipping it is why so many brands stall at a ceiling they cannot explain. The purpose here is not growth. It is to make the thing that worked in launch repeatable, predictable, and boring. You are turning a lucky first quarter into a reliable machine.

This means tightening your replenishment so you never stock out on a hero product. It means building the support and returns process that keeps account health green under more volume. It means knowing your real contribution margin per product after every fee, so that when you do scale spend, you are scaling something profitable rather than amplifying a leak. The work is unglamorous and it is the foundation everything later stands on.

What the stabilize gate actually measures

The gate out of stabilize is a capacity test, not a revenue test. Can your ops team handle double the current order volume without a drop in dispatch times. Can your cash cycle fund the inventory that scaling demand will require. Can you replenish your top products without a stockout for sixty straight days. If any answer is no, you stay in stabilize and close the gap. Pushing to scale on top of an unstable base just breaks the base faster.

Phase three: scale, months eight to twelve

Only now do you earn the right to pull the levers that look like growth. Wider keyword coverage and higher ad budgets, because the listings convert and the margin holds. A second marketplace, because the playbook is proven and documented. A broader catalogue, because the ops machine can absorb the added complexity. Scale is the easy part when the first two phases were done honestly. It is a disaster when they were skipped.

The reason scaling breaks brands is almost never demand. It is that revenue grows faster than the operation underneath it. Orders outrun fulfilment, returns outrun the support team, and inventory outruns cash. The roadmap protects you by tying every scale move to a capacity gate, so you never load more weight than the system can carry. We go deep on this failure mode in scaling from one crore to ten on marketplaces without breaking ops, because the leap is an operations problem long before it is a marketing one.

  • Scale spend only on listings with a proven conversion rate and a margin that survives the higher cost of traffic.
  • Add a marketplace only after the first one runs without daily firefighting.
  • Expand the catalogue at a pace your replenishment and cash cycle can actually fund.
  • Treat every capacity gate as a hard stop, not a suggestion you can override when you feel optimistic.

The roadmap is a sequence, not a calendar

The dates above are a guide, not a promise. Some brands clear the launch gate in six weeks. Some sit in stabilize for two quarters because their cash cycle needs the time. That is fine. The roadmap is defined by the gates, not the months. A brand that hits month eight without clearing the stabilize gate does not get to scale just because the calendar says so. It stays put until the capacity is real.

This is the part that founders find hard, because it asks them to delay the satisfying part. But the brands that respect the sequence are the ones still growing in year two, while the hockey-stick brands are untangling a backlog of returns and a suspended listing. A roadmap that survives contact with reality is not the most ambitious one in the room. It is the one that refuses to grow faster than it can carry. That patience is the whole edge.

What changed recently, and why it makes the sequence matter more

The channel mix a 2026 roadmap has to plan around is not the one most decks were built on. Quick commerce has gone from a side bet to a load-bearing channel, and that shifts where the capacity gates bite. Inc42’s D2C report frames the segment as a roughly eight billion dollar GMV opportunity on track to multiply by the end of the decade, with quick commerce projected to take a fifth to a quarter of D2C sales in metros by 2030, per Inc42. A roadmap that still treats Amazon and Flipkart as the only games in town is already a year behind.

It also got more expensive to be visible there. Industry reporting puts brand advertising on the big three quick commerce platforms at around 4,000 crore rupees in 2025, up sharply year on year, with projections near 6,000 crore for 2026, again per Inc42. On top of that the platforms have been layering on consumer-facing handling and platform fees, with Blinkit and Instamart adding charges through the year while Zepto rolled some back, as covered by Storyboard18. Both forces compress the margin you are supposed to defend in phase one, which is exactly why the stabilize gate around real per-SKU contribution is not optional. We unpack the math in quick commerce unit economics after platform fees.

The shelf is expanding fast too. Blinkit reported operating in the region of 2,200 dark stores by early 2026 and has signalled continued aggressive expansion toward roughly 3,000, with Zepto and Instamart building dense networks of their own, per Business Standard. More stores means more assortment slots and more places your replenishment and cash cycle have to feed without stocking out. That is a capacity problem before it is a demand problem, which is the whole thesis of this roadmap.

And the demand spikes you have to absorb are getting bigger. The 2025 festive season pushed marketplace GMV to record levels, with the opening week alone clearing tens of thousands of crore and Flipkart taking the larger share of the two big players, per Inc42. A brand that chases that spike before clearing its stabilize gate does not win the festival. It spends the next quarter cleaning up returns and account-health damage. None of this changes the roadmap. It raises the cost of skipping a phase.

We build these roadmaps as the core of our D2C & Marketplace Strategy Consulting, and we run them through Marketplace Account Management and Performance Marketing so the gates are enforced by people who own the outcome, not just the chart. A plan is only as good as the discipline behind it. Sequence the phases, gate the scale, and let the capacity decide the pace.

Nykaa Ad Placements: Buying Beauty Visibility That Pays Back

There is a particular kind of beauty brand that switches on Nykaa ads, watches the spend leave, and concludes the platform does not work. The visibility was real. The traffic arrived. It just landed on a product page with one flat image, a spec-sheet description, and four reviews, and it bounced. The ad did its job. The page had nothing to convert. This is the most common way to waste money on Nykaa, and it is entirely avoidable once you understand what the inventory is actually rewarding.

Nykaa ads are not Amazon’s search auction with a different logo. The platform is curated, editorial, and content-led, and its paid placements behave accordingly. They sit inside a discovery experience, not on top of pure search intent. Buy them with an Amazon reflex and you will overpay for the wrong audience. Buy them as an amplifier on assets that already convert and they compound. This is where Performance Marketing & Ads stops being a spend line and starts being leverage.

The inventory rewards content, not keywords

On a search-led marketplace, your ad wins when it captures loud intent. Someone types the product, you bid for the slot, the click is warm. Nykaa works differently because the shopper is often browsing, reading, and being guided by merchandising rather than hunting a specific SKU. We unpack why that distinction matters across the whole platform in how Nykaa rewards brands differently from Amazon, and it applies doubly to ads.

What this means in practice is simple. The placement delivers a browser to your page, not a buyer with intent. That browser has to be convinced on arrival. If the listing carries shade-accurate imagery, ingredient logic, a clear routine, and reviews seeded from real trial, the ad converts. If the listing is thin, you have paid full price to introduce a skeptical shopper to a page that gives them no reason to stay. The ad is only as good as the page it points at.

On Nykaa you are not buying clicks. You are buying the chance to put a content-rich page in front of someone who was browsing. Skip the content and you are buying a bounce.

Where the placements actually live

Nykaa’s paid inventory is broader than sponsored product slots, and treating it as one undifferentiated bucket is how budgets get blunt. The placements behave like different instruments, and an operator funds them for different jobs.

  • Sponsored product and search placements: the closest thing to intent capture. Useful when a shopper already knows the category, but a smaller slice of Nykaa traffic than on a pure search marketplace.
  • Banner and category-page placements: visibility inside the browse experience, where most discovery happens. These reward a strong creative and a page worth landing on, not a clever keyword.
  • Brand and edit features: proximity to Nykaa’s own editorial, which is where the platform’s audience is most receptive. These are partly earned through being a brand merchandising wants to feature, not purely bought.
  • Offer-window amplification: paid push timed to the platform’s big sale events, when buying intent is already elevated and trial-led purchasing peaks.

Spread spend evenly across these without a thesis and you learn nothing. Concentrate it behind the placements that match where your buyer actually decides, and you get signal you can compound on.

Pair every placement with an editorial asset

This is the rule that separates Nykaa ad spend that pays back from spend that evaporates. Behind every placement should sit content built to convert a browser, not a spec sheet built to be found. That is real photography, swatches, written routines, ingredient stories, and a review base seeded from sampling. We go deep on what that looks like in beauty content that converts on Nykaa and beyond, and the sequencing is not optional. Content first, paid second.

The reason is mechanical. A thin page does not just convert worse. It tells you nothing about whether the placement works, because you cannot separate a weak ad from a weak landing experience. Fund the content, prove the page converts organically, then put paid behind the assets that are already earning. Now your ad spend is amplifying a known winner instead of gambling on a page you never tested. This is the discipline at the centre of Performance Marketing & Ads and D2C & Marketplace Strategy Consulting when the two are run as one motion rather than two budgets.

How to measure whether it paid back

Beauty pays back on the second purchase, not the first, and ad measurement that ignores this will mislead you on Nykaa more than almost anywhere. A trial-led category means the value of a first buyer is mostly latent. Judge a placement on click cost alone and you will kill the ones that are quietly building a repeat base.

  1. Measure cost per acquired buyer against repeat rate, not cost per click. A pricier acquisition that repeats beats a cheap one that does not.
  2. Read the landing page’s organic conversion before you scale paid behind it. If it does not convert without ads, more ads will not fix it.
  3. Separate intent placements from discovery placements in reporting. Blending them hides which mechanism is actually working.
  4. Weight offer-window performance for the trial it generates, not only the immediate margin, since first trial is how beauty buyers commit.

Run the numbers this way and the picture changes. Placements that looked expensive on a click basis often look efficient on a repeat-buyer basis, and the cheap-looking ones sometimes buy traffic that never comes back.

Where Nykaa sits in a wider ad budget

Most beauty brands are not only on Nykaa, and the temptation is to run one ad playbook across every platform. That is a reliable way to underperform everywhere. Each marketplace has its own auction logic, its own buyer behaviour, and its own definition of a good placement. We lay out how to hold one budget across very different rule sets in performance marketing across marketplaces. Nykaa’s slice of that budget earns its place on content and editorial proximity, not on keyword volume.

The practical consequence is that you cannot copy your Amazon bid strategy onto Nykaa and expect it to translate. The same rupee buys a different thing. On one platform it captures intent. On the other it amplifies discovery. Treating them as interchangeable is how cross-platform brands quietly waste a third of their spend.

What changed recently

The platform’s own numbers confirm why ads here are a serious channel rather than a side bet. Nykaa crossed the one billion dollar revenue mark across fiscal 2026, with Q4 FY26 gross merchandise value up 28 percent to about Rs 5,241 crore and net profit up 313 percent year on year, according to Storyboard18. A platform growing repeat beauty demand at that pace is a platform whose ad inventory is getting more valuable, not less.

Two operator-relevant shifts sit underneath those numbers. First, Nykaa is deliberately broadening its advertiser base. CEO Anchit Nayar has said no single brand contributes double-digit ad revenue, with thousands of brands now buying placements, and that the platform has built more participation opportunities across categories while sharing real-time campaign analytics through AI, per the same Storyboard18 report. For a challenger brand that reads as access: the auction is no longer reserved for the biggest spenders, and the reporting is finally good enough to separate a working placement from a wasted one.

Second, discovery itself is moving. Nykaa has signed a multi-year deal with OpenAI to put its beauty and fashion storefronts inside ChatGPT, letting shoppers ask for a skincare fix or an outfit and get product recommendations without opening the app, as reported by Storyboard18. Financial terms and a launch date were not disclosed, and it is early. But the direction confirms the thesis of this whole piece. As discovery shifts from keyword search toward conversational recommendation, the brands that win are the ones whose pages and product data are rich enough to be recommended, not just found. Content remains the asset. The places it gets surfaced are only multiplying.

Get this right before you spend

Almost every decision that determines whether your Nykaa ads pay back is made before the first placement goes live. The content you prepared, the assortment you submitted, and the trial budget you committed all shape how the platform receives you and how your pages convert. We cover the approvals and the buffers to plan for in onboarding a beauty brand to Nykaa, and the order is the whole point. Build the content-rich pages, prove they convert, then buy visibility to pour onto them.

Do it backwards, ads first and content last, and Nykaa will feel like a platform that does not work. It works. It just refuses to subsidise a thin page with paid traffic. Buy beauty visibility on Nykaa the way the platform actually rewards it, behind editorial assets that earn the click, and the spend pays back through the repeat-buying customers that beauty was always going to be about.

Review Generation That Stays Inside Marketplace Rules

Every brand new to Amazon India asks the same question in the first month. How do we get reviews faster. The honest answer disappoints them. The methods that move the number quickly are the ones that get you suspended, and the methods that survive an audit move the number slowly and steadily. Review generation is not a growth hack. It is an operational discipline, and the brands that treat it like a discipline are the ones still standing when the incentivised-review crowd gets purged.

Here is the line that matters. Amazon does not ban you for having few reviews. It bans you for the way you got them. A listing with forty honest reviews outlives a listing with four hundred bought ones, because the four hundred are a liability waiting for a sweep. Once you accept that the rules are the constraint you build inside, the work becomes clear. It is inserts, follow-ups, and Vine. It is not gift cards in exchange for five stars.

The bright line you do not cross

Start with what is forbidden, because most brands get this wrong without realising. Amazon prohibits any review that is incentivised, compensated, or solicited in exchange for a benefit. That includes the obvious schemes and the clever ones. A free product for an honest review is banned. A refund after a five-star review is banned. A WhatsApp group where buyers post screenshots for cashback is banned. Inserting a card that asks only for positive reviews, or that funnels unhappy buyers to email while pushing happy ones to the review page, is banned.

The pattern Amazon polices is selection and reward. The moment you reward a review, or filter who gets asked based on how they feel, you have crossed the line. Detection is better than sellers assume. Review velocity that spikes unnaturally, reviewer accounts that cluster across your catalogue, and language patterns all feed the same machine that watches your account health metrics. A review purge does not just delete the reviews. It can take the listing and sometimes the account with it.

A bought review is a loan against your account. The reviews look like an asset until the day Amazon calls the debt, and it always calls the debt.

Inserts that ask, not bribe

The package insert is the most underused compliant tool on Indian marketplaces. Done right, it is a small card inside the box that thanks the buyer, helps them use the product, and points them to leave a review. Done wrong, it is a suspension trigger. The difference is entirely in the ask.

A compliant insert asks every buyer for an honest review, with no condition and no filter. It does not offer anything in return. It does not say the word positive. It does not route unhappy customers away from the review button. What it can do is reduce the friction that stops happy buyers from bothering. Most satisfied customers never review because reviewing is a chore, not because they are unwilling. The insert closes that gap.

  • Lead with usefulness. A setup tip, a care instruction, or a quick-start guide earns the buyer’s attention before you ask for anything.
  • Ask neutrally. Invite an honest review of their experience. Never qualify it with positive, five-star, or any sentiment word.
  • Make it one tap. A short URL or QR code to the review page removes the navigation tax that kills most intent.
  • Offer support, separately. Give a real contact for problems, but do not make it a detour that only the unhappy take. Show it to everyone equally.
  • Never mention a reward. No discount, no entry, no free anything tied to the review. That single line voids the whole exercise.

The card carries your brand, so treat it like a touchpoint, not a receipt. The same buyers reading the insert are the ones who will shoot the unboxing clips that feed a real UGC and review strategy, so the insert and your content engine should speak with one voice.

Amazon Vine, the only paid path that is allowed

Vine is the exception that proves the rule. It is the one programme where Amazon itself supplies reviewers in exchange for free product, and it is fully compliant because Amazon controls the selection and the disclosure. Enrolled brands submit a limited number of units, Vine Voices receive them, and the reviews post with a clear Vine label so buyers know the unit was given.

Vine is built for the cold-start problem. A new listing with zero reviews converts poorly no matter how good the product is, because buyers will not be the first to risk it. Vine seeds the first credible reviews so the flywheel can start. Two things to hold in mind. Vine reviews are honest, not guaranteed positive, so a weak product gets honestly criticised in public. And Vine is capped, so it is a starter, not a scaling engine. It gets you from zero to a baseline. After that, inserts and organic velocity carry the load.

We treat Vine as a launch lever inside Marketplace Account Management, sequenced with the listing work so the reviews land on a page that is already built to convert. Seeding reviews onto a weak listing wastes the allocation. The Vine units should hit a page where the imagery and copy already earn the click, which is why we run review seeding and listing conversion work as one motion, not two.

The follow-up loop most brands skip

Amazon’s own Request a Review button is the most boring and most compliant tool available, and almost nobody uses it consistently. It sends a templated, Amazon-controlled message asking the buyer to review. You cannot edit it, which is exactly why it is safe. There is no room to incentivise or filter because Amazon writes the words.

The leverage is in cadence, not cleverness. A brand that triggers a review request on every eligible order, on a fixed rhythm, after the delivery window has comfortably closed, will out-accumulate a brand that does it ad hoc. This is operations, not marketing. It belongs in the same weekly cadence as your health-metric review, run by the same person who watches dispatch and defect rates. Velocity from this loop is slow and durable, the opposite of a bought spike, and it never shows up as a red flag because Amazon generated every message itself.

What you do with the reviews you earn

Generation is half the system. The other half is response, and it is where most brands quietly leak trust. Every review, good or bad, is a public signal to the next thousand shoppers. A thoughtful reply to a critical review often persuades the reader more than the original complaint repels them, because it shows a brand that shows up. We have a full method for this in our piece on responding to negative reviews without making it worse, and it pairs directly with everything here.

What changed recently

Two shifts in the last year make the compliant path less of a virtue and more of a necessity. The first is enforcement scale. In its inaugural Trustworthy Shopping Experience Report, Amazon said it blocked hundreds of millions of suspected fake reviews and seized more than 15 million counterfeit products globally in 2025, shut down over 100 websites tied to review fraud, and pushed more than 40 fake-review brokers to cease operations, per Business Standard. Crucially, Amazon also expanded its Counterfeit Crimes Unit into India, signalling tighter coordination with brands, sellers, and law enforcement on this exact problem. The broker who promises you a hundred reviews this week is now selling you into a litigation pipeline, not a shortcut.

The second shift is regulatory. India’s standard for online reviews, IS 19000:2022, was voluntary at launch, but the Department of Consumer Affairs has been moving to make it mandatory after the voluntary phase produced limited results. Major platforms including Amazon and Flipkart have backed the move, as reported by Business Standard. The standard already bars paid and incentivised reviews and blocks the suppression of negative ones, and a consumer survey cited by LocalCircles found roughly eight in ten users want these rules mandated. Read the two together and the direction is obvious. The methods that look like a shortcut today are the ones a stricter regime will treat as an unfair trade practice tomorrow.

The mindset shift is the whole game. Stop asking how to get reviews fast. Start asking how to deserve them and how to make leaving one effortless. Build the insert, enrol Vine at launch, run the Request a Review loop on a cadence, and respond to everything. It is slower than buying stars. It is also the only version that is still working a year from now, which is the only timeline a serious brand should plan on.

Performance Marketing Across Marketplaces: One Budget, Different Rules

Most brands we inherit run their marketplace ads as if there were one platform with three logos. The same keyword logic, the same bid settings, the same daily caps, cloned from Amazon and pasted into Flipkart and quick commerce because Amazon is where the team learned the craft. It feels efficient. It is actually three quiet leaks running at once. Amazon, Flipkart and the quick commerce apps reward completely different behaviors, and a budget that ignores that difference pays full price on every platform while only winning on one.

The hard part is that the budget really is one number. A founder approves a monthly ad spend and expects it to work across the channels the brand sells on. So the job is not to run three disconnected accounts. It is to allocate one pool intelligently, then run platform-native tactics inside each slice. Unified at the top, specific underneath. That is the whole discipline, and almost nobody does it.

Why one playbook breaks across three platforms

Amazon India is an auction with deep keyword data and a mature relevance engine. It rewards patience, clean campaign structure, and bids that reflect what a search term is actually worth. You can gather weeks of conversion data and prune with confidence. The system is built to be read.

Flipkart is also an auction, but the relevance signals, the category dynamics, and the buyer behaviour are different enough that a bid which wins on Amazon routinely overpays or underdelivers here. The biggest mistake is assuming your Amazon bidding instinct transfers. It does not, and we wrote a whole piece on exactly why your Amazon bidding logic underperforms on Flipkart PLAs. The placement economics, the competitive density, and the conversion patterns all differ.

Quick commerce is not even the same kind of system. There is no long tail of search to harvest. Shelf space is a few screens, assortment is decided dark store by dark store, and visibility is bought against a tiny, contested surface. Buying visibility on Blinkit is closer to retail trade marketing than to search advertising. You are paying for position in a coded shelf, not for intercepting intent across a million queries.

One budget can be unified. The tactics underneath it cannot. Treating Amazon, Flipkart and quick commerce as one auction is the most expensive shortcut in Indian marketplace advertising.

The behaviors each platform actually rewards

Strip away the dashboards and each channel is paying you to do a specific thing. Get the thing right and the spend compounds. Get it wrong and you fund the platform’s revenue with none of your own.

  • Amazon rewards structure and data discipline. Tight keyword segmentation, the right ad type for the funnel stage, and bids that evolve as conversion data arrives. The brands that win here are the ones that read the data before they cut.
  • Flipkart rewards native bid logic and category awareness. The PLA auction behaves on its own terms, so you tune to Flipkart’s signals rather than importing Amazon’s. Underbidding loses placement you could afford, overbidding burns margin on traffic that converts worse.
  • Quick commerce rewards assortment and shelf position over keyword cleverness. Visibility is the product. If you are not in the dark store and not on the first screen, no bid saves you. Spend follows availability, not search volume.

Notice these are not three flavours of the same skill. They are three different skills that happen to share a line item on the budget.

Allocating one budget without flattening it

The allocation question is where most brands either freeze or default to splitting evenly, which is the same mistake as splitting evenly between ad types on a single platform. Equal is not the same as right. The split should follow where the channel sits in your business and how efficiently it converts spend, not a fairness instinct.

We start with three inputs before a single rupee moves. Where is the demand. Where is the margin. Where is the efficiency. A platform that drives volume but thin margin gets funded differently from one that drives fewer orders at healthy contribution. Quick commerce in particular can look expensive on a pure cost basis while being the only place a category is actually growing, so the allocation has to weigh strategic position, not just last-click return.

A working sequence we use

  1. Set the unified budget at the top, then assign each platform a job. Defend, grow, or test. The job decides the rules, not the other way round.
  2. Fund the platform where your data is cleanest and your conversion is proven first. Usually that is Amazon for established sellers. Let it carry the largest slice early.
  3. Give Flipkart its own native targets and bid logic. Never benchmark it against Amazon’s ACoS as if the auctions were comparable. They are not.
  4. Treat quick commerce spend as a function of assortment and shelf presence. Pair the ad budget with availability, because paying for visibility on a SKU that is out of stock in the dark store is pure waste.
  5. Rebalance monthly off real numbers. The split is never fixed. It moves as each channel proves what it can convert.

The bid mechanics themselves differ by platform too, and the choice between fixed, dynamic and rule-based bidding is not one global setting. It depends on the campaign’s job and how much clean data you already have. We break that down in our note on when to use fixed, dynamic and rule-based bids, and the short version is that you pick the bid type per platform and per campaign maturity, not once for the whole account.

The reporting trap that hides the leaks

The reason cross-marketplace waste survives so long is that blended reporting hides it. A founder sees one combined ACoS across all channels, it looks acceptable, and nobody digs. Underneath, Amazon is subsidising a Flipkart account that is quietly overbidding and a quick commerce spend that is paying for invisible SKUs. The average looks fine. The components are bleeding.

So the first thing we do on any multichannel account is break the numbers apart. Per platform efficiency, per platform job, per platform trend. Only then can you tell whether each slice of the unified budget is earning its place or coasting on the blended average. A healthy total made of one strong channel and two weak ones is not healthy. It is one channel carrying the bill.

Where the platforms do talk to each other

None of this means the channels are islands. They share a brand, a margin structure, and often a shopper who sees you in more than one place. The keywords that convert on Amazon inform what you emphasise on Flipkart listings. The hero SKUs that move on quick commerce tell you what to push hardest everywhere. The data is shared intelligence even when the tactics are separate.

That is the balance to hold. Unified strategy and shared learning at the top. Platform-native execution underneath. The brands that get this right stop asking which marketplace is best and start asking what each one is for. One is defending share. One is growing into a younger audience. One is buying physical-feeling shelf presence in the fastest channel in the country. Different jobs, funded deliberately, run by their own rules.

What changed recently

The reason this discipline matters more in 2026 than it did two years ago is that the ad surfaces themselves have become a primary revenue engine for the platforms, which means they are getting deliberately better at extracting your budget. In FY25 Amazon India’s ad sales rose about 25 percent to roughly 8,342 crore rupees and Flipkart booked around 6,310 crore, together about 14,652 crore in commerce advertising, per Exchange4media. The same report notes advertising is now close to 28 percent of Amazon India’s operating base, and for Flipkart marketplace services more than doubled on the back of brand promotions. When ads are a quarter to a third of a platform’s revenue, the auction is no longer a neutral utility. It is the product they are selling you.

Quick commerce is where the curve is steepest. A Datum Intelligence estimate cited by Storyboard18 puts Blinkit, Zepto and Instamart combined ad revenue at close to 4,900 crore rupees for 2026, with total quick commerce ad spend running between 5,000 and 6,000 crore annually. That is a young surface monetising fast. It rewards early movers who understand the shelf, and it punishes brands that bolt an Amazon keyword strategy onto a channel that does not run on keywords. The lesson is the one this piece opened with, now with numbers behind it. Three different games, three different rates of change, one budget that has to keep up with all of them.

One budget, run like three operators

The takeaway is not complicated, but it is uncomfortable for any team that built its competence on a single platform. The Amazon playbook is a great Amazon playbook. It is a mediocre Flipkart playbook and an actively wrong quick commerce playbook. A unified budget does not mean uniform tactics. It means one disciplined allocation feeding three native games.

This is the core of how our Performance Marketing & Ads work runs across channels, allocating a single budget against per-platform jobs and reading each slice on its own terms rather than a flattering blended average. It is reinforced by the listing and shelf discipline of our Marketplace Performance practice, because the cleanest bid in the world loses money if the page underneath it does not convert or the dark store does not have stock. Run one budget. Run it like three different operators. The brands that separate the rules win all three channels instead of paying full freight to win one.

Flipkart PLA Strategy: Why Your Amazon Bidding Logic Underperforms Here

The most expensive mistake on Flipkart Product Listing Ads is treating the platform as Amazon with a different logo. Teams export their Amazon keyword bids, their target ACoS, their dayparting rules, and paste the whole framework into Flipkart Ads. Then they watch spend climb and orders stay flat, and they blame Flipkart traffic quality. The traffic is fine. The bidding logic is wrong, because it was built for a different auction with different rules. Flipkart does not reward the behaviour Amazon rewards, and the gaps are specific enough to budget around.

We run paid media across both marketplaces, and the brands that win do not have a clever trick. They have two separate bid models. They stopped pretending one set of numbers transfers. Here is where Flipkart actually diverges, and what to rebuild.

The auction is not the same shape

Amazon’s Sponsored Products auction has trained a generation of operators to think in tight keyword targets with granular bid control per term. You bid a keyword, you see its placement, you adjust. Flipkart PLA leans more on the category and product context than on the long tail of exact-match keywords. Your bid competes inside a placement logic that weighs your listing’s own conversion signals heavily, not just your willingness to pay.

The practical effect is that a high bid on a weak listing does much less on Flipkart than the same high bid would buy you on Amazon. Flipkart is quicker to let a strong organic performer ride a modest bid past a paying competitor with a worse listing. If your Amazon instinct is to outbid your way into a placement, Flipkart will charge you for the attempt and still hand the slot to the better-converting product.

On Amazon you can sometimes buy your way past a weak listing. On Flipkart the listing has a vote, and it often outvotes your bid.

Conversion history is weighted differently

Both platforms care about conversion. Flipkart cares earlier and harder for ad placement. A product with thin order history and a soft rating gets less mileage from aggressive bidding here than the same product would on Amazon, where a fat bid can brute-force impressions while the listing matures.

This changes launch sequencing. On Flipkart you do not open with maximum bids on a cold listing. You get the fundamentals right first, because the ad engine compounds whatever conversion signal you already have. The order is review velocity, then price competitiveness, then bids. Pour budget into a listing that has not earned its conversion rate and Flipkart spends it inefficiently and learns slowly. This is the same discipline we apply when we decide when to use fixed, dynamic, and rule-based bids, just tuned to a platform that punishes premature aggression more sharply.

Placement inventory and seasonality behave differently

Flipkart’s demand is more event-shaped than Amazon’s steady baseline. The platform concentrates enormous volume into tentpole sale windows, and the auction during those windows behaves like a different marketplace entirely. Bids that are comfortable in a normal week get overrun the moment a sale event opens, because every competitor floods in at once and the cost floor jumps.

If you carry your steady-state Amazon bidding cadence into a Flipkart event, one of two things happens. Either you underbid and vanish from the placements that matter most in the only week that matters most, or you leave your normal caps in place and get your whole budget eaten in two days. Neither is an accident of the platform. It is a failure to plan for an auction that breathes in spikes. We go deep on the run-up in our guide to planning inventory and ads months ahead of Big Billion Days, because the bidding decision is made weeks before the event, not during it.

What to actually rebuild

Stop porting and start rebuilding. The transferable thing between Amazon and Flipkart is your discipline, not your numbers. Here is the short list of what needs its own Flipkart version:

  • Bid baselines: set them from Flipkart’s own placement costs, not your Amazon CPCs. They are different auctions with different competitor sets.
  • Target efficiency: hold a separate ACoS or ROAS target per platform. Margins differ because Flipkart’s fee and commission structure differs, so the break-even bid differs.
  • Listing readiness gate: do not scale bids on a listing that has not earned a conversion rate. On Flipkart this gate is stricter, so enforce it before spend, not after.
  • Event bid plan: maintain a sale-window bid ladder that is separate from your everyday bids, with caps that assume the cost floor jumps.
  • Keyword versus category split: lean more on category and product targeting on Flipkart, less on the granular exact-match keyword sprawl that Amazon rewards.

The budget is one pool, the rules are not

Most brands we work with run a single marketplace media budget and then make the mistake of governing it with a single set of rules. The budget can absolutely be shared. The bid logic, the efficiency targets, and the pacing cannot. A rupee spent on Flipkart and a rupee spent on Amazon buy different things, against different auctions, at different times of the season. Treating them as interchangeable is how money leaks quietly while the dashboard looks busy.

This is the core argument in our piece on running one budget across marketplaces with different rules. The allocation decision is shared. The execution is platform-specific. The teams that conflate the two end up optimising one platform’s logic onto another and underperforming on both.

What changed recently

The case for treating Flipkart Ads as its own discipline has only hardened, because the platform now treats advertising as a primary profit engine rather than a side menu. Flipkart’s ad income rose 27% to Rs 6,317 crore in FY25 and now contributes roughly 31% of its marketplace revenue, per Storyboard18. Across Amazon, Flipkart and Myntra together, e-commerce advertising revenue crossed Rs 15,573 crore in FY25, up about 26% year on year, according to IBEF. When a marketplace earns nearly a third of its money from your bids, the auction is being engineered for the platform’s margin, not yours. That is a structural reason the placement logic and cost floors keep drifting away from anything your Amazon model would predict.

The seasonality point sharpened too. For Big Billion Days 2025, Flipkart did not dramatically lift total spend so much as reallocate it, doubling Connected TV investment and pushing influencer-led commerce from negligible to central, as reported by Storyboard18. The platform is funnelling more upper-funnel demand into the same event windows where your PLA cost floor already jumps. The operator read is unchanged but more urgent: build the Flipkart event bid ladder deliberately, because the traffic surge you are bidding into is being manufactured harder each festive season.

Where an operator earns the difference

None of this is exotic. It is the discipline of refusing to assume that a number which worked in one auction works in another. The Amazon playbook is good. It is just an Amazon playbook. On Flipkart you need a Flipkart model, built from Flipkart’s placement costs, conversion weighting, and event rhythm, run alongside Amazon rather than copied from it.

That is the heart of our Performance Marketing & Ads work across Indian marketplaces. We build and govern the per-platform bid models, the shared-budget allocation logic, and the event ladders that keep spend efficient when the auction spikes. The brands that grow on Flipkart are not the ones bidding hardest. They are the ones who stopped pretending it was Amazon and rebuilt the logic for the platform they were actually buying on.

Flipkart Big Billion Days: Planning Inventory and Ads Months Ahead

Every year the same pattern repeats. In late August, brands wake up to the fact that Big Billion Days is weeks away. They scramble to top up stock, throw together a discount sheet, and pile budget into ads the day the event opens. Then they spend the post-mortem complaining about stockouts, thin margins, and ad costs that doubled overnight. None of that was bad luck. It was the predictable result of planning a tentpole event as if it were a surprise. BBD is on the calendar every year. The winners treat it like the fixed deadline it is, and they start a quarter early.

We run marketplace media and operations for brands across India, and the gap between a good BBD and a bad one is almost never execution during the event. It is everything decided before it. The teams that grow lock inventory and ad budgets months ahead. The teams that struggle decide in real time, which is exactly when the platform’s economics work against them.

The decision window closes long before the sale

The instinct is to think of BBD as an event you run. It is more useful to think of it as an event you commit to. By the time the sale opens, every meaningful lever is already set. Your stock is in the warehouse or it is not. Your price is locked into the platform’s deal structure or it is not. Your ad budget and bid ladder are configured or you are improvising against competitors who configured theirs in June.

This is why last-minute prep feels so expensive. You are making decisions in the one window where you have the least leverage and the platform has the most. Inbound logistics are congested, the ad auction is at its cost floor, and discount commitments are non-negotiable. Move those same decisions back a quarter and the leverage flips. You ship inventory before the rush, you model your discount against margin calmly, and you set bids before every competitor floods the auction.

BBD is not a week you survive. It is a quarter you plan, compressed into a few days of execution.

Inventory is the bet you cannot unmake mid-event

Of everything, inventory is the least forgiving. You cannot conjure stock during the sale. If you sell out on day two of a five-day event, you have handed the back half of your demand to a competitor and trained the algorithm that your listing goes dark when traffic peaks. If you overstock, you carry the cost of dead inventory into a slow Q4 and end up fire-selling to clear it.

The hard part is that BBD demand is spiky and non-linear. Your steady-state sell-through tells you almost nothing about a sale-week peak that can run many times your normal volume on the hero SKUs and barely move the long tail. This is where most forecasts break. We treat sale-event forecasting as its own discipline, separate from baseline planning, because the math is different. Our approach to forecasting demand when it is spiky rather than smooth is built precisely for these windows, where a single week distorts the whole quarter.

The planning order we use is simple to state and hard to do well:

  • Rank SKUs by event-week potential, not annual volume. The products that win BBD are not always your everyday bestsellers. Deal-friendly price points and gifting demand reshuffle the ranking.
  • Set a depth target per hero SKU with a deliberate buffer. Stocking out at the peak costs more than the carry on a modest overstock. Bias toward not going dark.
  • Book inbound logistics early. Warehousing and fulfilment slots get scarce as the event nears. Late inventory that misses the cutoff is the same as no inventory.
  • Reserve a replenishment plan you can actually trigger. A mid-event top-up only helps if the lead time fits inside the sale window, which usually means it must already be in transit.

Discount is a margin decision, made in advance

The platform wants your deepest possible discount, because depth drives the visibility it sells. Your job is to protect margin while still earning placement. That tension cannot be resolved in the panic of event week. You need to know your floor before you commit, and you need to know which SKUs you are willing to run thin as traffic drivers versus which ones carry the margin.

That is a pricing architecture, not a spreadsheet you fill in the night before. We lay out how to defend the bottom line when discounting is mandatory in our piece on protecting margin when everyone around you discounts. The short version is that a discount you modelled in advance is a strategy, and a discount you agreed to under deadline pressure is a leak.

Ad budgets get set in the calm, not the storm

The BBD auction does not behave like a normal week. Every competitor floods in at the same moment, the cost floor jumps, and bids that were comfortable in August get overrun on day one. If you carry your steady-state bidding into the event, you either underbid and vanish from the placements that matter in the only week that matters, or you leave normal caps in place and watch your entire budget evaporate in forty-eight hours.

Neither outcome is a platform problem. It is a planning problem. The bid decision is made weeks before the sale opens, with a separate event bid ladder, separate efficiency targets, and caps that assume the floor jumps. Flipkart’s auction also weighs your listing’s own conversion signal heavily, which means a strong listing earns placement more cheaply than a competitor brute-forcing it with budget. That dynamic is the core of our Flipkart PLA bidding logic, and it matters more during BBD, not less, because the cost of every inefficient impression multiplies when the floor is high.

Set your event ad plan in the calm of the prior quarter. Decide your daily caps, your hero-SKU bid priority, and the point at which you stop chasing unprofitable impressions. Then during the event you are governing a plan, not inventing one while spend runs hot.

The whole thing is one rehearsed motion

The reason early planning works is that BBD is not really three separate problems. Inventory, pricing, and ads are one system. Your discount depth determines your sell-through, which determines the inventory depth you need, which determines how hard you can afford to bid before margin disappears. Decide any one of these in isolation, at the last minute, and the other two go wrong. Decide them together, a quarter ahead, and they reinforce each other.

This is also why BBD prep rhymes with the rest of the festive calendar. The same operators who plan Flipkart’s tentpole well tend to run a clean Great Indian Festival prep plan too, because the muscle is identical. Forecast the spike, model the margin, set the auction plan, ship the stock early. The platform changes. The discipline does not.

What changed recently

The 2025 festive run reset what a hero quarter looks like, and it reset it in ways you have to plan for, not react to. The GST 2.0 reform landed right before the season, cutting rates on several appliance and mid-priced categories from 28 percent to 18 percent and making many products noticeably cheaper at checkout. That is not a footnote. According to Business Standard, total festive e-commerce sales were projected to grow about 27 percent year on year past ₹1.2 trillion, with the first week alone generating roughly ₹60,700 crore in GMV. When a tax cut pulls demand forward and concentrates it, the brands that pre-modelled their depth against the new price points captured it. The ones still treating discount as a deadline decision left margin on the table or went dark on the hero SKUs.

The demand also moved deeper into the country and faster down the delivery clock. Independent forecaster Redseer called it the strongest festive period in five years, with GMV set to cross ₹1.15 trillion and tier-II and tier-III cities leading the growth. On the speed side, Flipkart Minutes pushed quick commerce into the festive event itself, with premium electronics, not just groceries, emerging as a quick-commerce driver during the sale, as Business Standard reported. The planning consequence is concrete. If your category is now winnable in minutes, your inventory has to sit in the right dark stores and city pools before the sale, not just in a central warehouse. If you sell where tier-II demand is quadrupling, your depth targets and your ad geo-priorities should reflect that, not last year’s metro-weighted mix. None of this changes the discipline. It just raises the cost of skipping it.

What an operator actually does about it

The brands that grow on BBD are not the ones bidding hardest or discounting deepest during the event. They are the ones who made the hard calls in June and spent September simply executing a plan they already trusted. That is the heart of how we run Performance Marketing & Ads for Indian marketplaces. We build the event ad ladders, the per-SKU inventory targets, and the margin-aware discount architecture months before the sale, so that when the auction spikes and the traffic floods, our brands are governing a rehearsed motion instead of feeding the platform’s fees in a panic. Last-minute prep is not cheaper or faster. It is just more expensive, paid out in stockouts, eroded margin, and ad spend that buys less than it should.

PPC Bid Strategy: When to Use Fixed, Dynamic and Rule-Based Bids

Open a fresh campaign in any Indian marketplace ad console and the platform will gently steer you toward dynamic bidding. Up and down. It sounds like the responsible default, the option that lets a smart system flex your bid in real time toward conversions. For a campaign that has been running for months, on keywords you understand, it often is. For a campaign launched yesterday, on keywords that have converted exactly zero times, it is the most expensive setting you can choose.

That is the mistake we see most often. Brands pick a bid type once, at setup, based on which one sounds most advanced, and never revisit it. The better way to think about it is that bid type should follow campaign maturity. What you know about a keyword decides how much control you hand to the algorithm, not the other way round.

The three bid types, in plain terms

Strip away the platform names and there are three things you can do with a bid.

  • Fixed bids. You set a number and it stays there. The platform does not raise it on a click it thinks will convert, and does not lower it on one it doubts. You pay your bid, full stop. Boring, predictable, and exactly what you want when you have no data to trust yet.
  • Dynamic bids. The platform adjusts your bid up, down, or both, in real time, based on its own estimate of how likely that impression is to convert. It is using conversion signal you do not see. Powerful when that signal is real, dangerous when it is guessing.
  • Rule-based bids. You define the logic. Raise bids when ACoS sits below a target, cut them when spend runs ahead of sales, push harder on the placements that perform. The intelligence is yours, applied automatically, on your terms.

None of these is better than the others. Each is correct at a different point in a campaign’s life. The skill is knowing where you are.

Why dynamic bids punish new keywords

A dynamic bid is only as good as the conversion data feeding it. When you launch a brand-new keyword, the platform has no history for that term against your specific listing. So when it decides to raise your bid on an impression it judges likely to convert, that judgement is built on thin air. It is extrapolating from category averages and broad patterns, not from your actual performance.

The result is predictable. The system spends aggressively on clicks it has flagged as promising, those clicks do not convert because the prediction was a guess, and your spend balloons while your data stays too noisy to learn from. You have paid premium prices to discover what a fixed bid would have told you more cheaply. This is the trap behind defaulting to dynamic on day one. You are letting an algorithm bet your money on keywords neither of you has tested.

Dynamic bidding does not create conversion signal. It spends faster against the signal you already have. On a new keyword, that signal is noise.

This matters more every quarter, because clicks are not getting cheaper. One Amazon India seller-tooling analysis pegs CPCs as having climbed roughly 40 to 80 percent between 2023 and 2026 as more sellers crowded the auctions, with top-of-search positions carrying a further 20 to 40 percent premium (eVanik). When the auction floor keeps rising, the cost of letting an algorithm guess on your behalf rises with it. A fixed bid keeps that tuition affordable instead of letting the platform inflate the bill.

The maturity ladder

Think of bid type as three rungs you climb as a keyword earns trust.

Rung one: launch on fixed bids

A new campaign, a new keyword, no conversion history. Set a fixed bid at a sensible level for the category and hold it. You are buying clean, evenly priced data. Because the bid never moves on the platform’s whim, every click costs what you expected, and the conversion rate you observe is honest. After a few weeks you will know which keywords convert, at what cost, and which are dead weight. That clarity is impossible to read through the noise of an aggressive dynamic bid.

If your platform only offers dynamic on a new campaign, the safest cousin of a fixed bid is dynamic set to down-only. It will never outbid you on an impression it doubts, which caps your downside while you learn. Use it as a fixed bid stand-in, not as the real thing.

Rung two: graduate proven keywords to dynamic

Now you have data. A keyword has converted consistently, its ACoS is in a range you can live with, and the platform finally has real signal for that term against your listing. This is when dynamic bidding earns its place. The algorithm is no longer guessing. It is flexing your bid on patterns it has actually learned, pushing harder on the impressions most likely to convert and easing off the rest. Move your winners here. Leave your unproven terms on fixed.

Rung three: layer rule-based control over the portfolio

Once you are running dozens of keywords across several campaigns, manual management stops scaling and pure dynamic bidding gives away too much control. Rule-based bidding is how you keep your hand on the wheel at volume. You encode the decisions you would make anyway. Cut bids on terms drifting above target ACoS, raise them on placements quietly outperforming, pause spend that runs ahead of sales. The platform executes your logic, not its own.

The metric that tells you when to climb

Knowing when a keyword is ready to graduate means watching the right number. ACoS alone will not tell you, because a keyword can post a flattering ACoS while contributing nothing to organic rank. The fuller picture comes from reading ad efficiency against the whole account, which is the case we make in ACoS versus TACoS. A keyword worth promoting to dynamic bidding is one with stable conversions, a defensible cost, and enough volume that the platform has genuine signal to work with. Thin, sporadic converters stay on fixed no matter how good a single week looked.

The same discipline shapes which campaign types you trust dynamic bids with at all. Tightly converting branded and bottom-funnel terms reach maturity fast. Broad, upper-funnel discovery terms stay volatile for longer and deserve a fixed bid well past the point you think they have settled.

Bid type is a campaign-structure decision, not a checkbox

This is why bid strategy cannot be set once and forgotten. Your account is always a mix of keywords at different maturities, which means it should be a mix of bid types at any given moment. Healthy structure looks like this.

  1. A fixed-bid layer for discovery. New terms, new match types, new product launches. This is your testing ground, priced to keep learning cheap.
  2. A dynamic-bid layer for proven performers. Graduated keywords with real history, where letting the algorithm flex genuinely lifts return.
  3. A rule-based layer for portfolio control. The guardrails that keep the whole account inside your targets without manual babysitting.

Keywords move up the rungs as they prove themselves and drop back down when performance decays. That movement is the actual work of bid management, and it is what separates an account that compounds from one that just spends. The structure that supports it also has to survive the differences between platforms, because the same keyword behaves differently on each, a point we expand on in our take on running performance across marketplaces on one budget.

What changed recently

Two shifts make this maturity discipline more urgent than it was a year ago. The first is cost. Amazon India CPCs have run up an estimated 40 to 80 percent since 2023, with beauty among the steepest categories (eVanik). When clicks cost that much more, a dynamic bid guessing on an unproven keyword is not a small leak. It is a structural drain. The case for learning cheaply on fixed bids first only gets stronger as the auction heats up.

The second is where the money is moving. Ad spend on the quick commerce big three, Blinkit, Zepto and Swiggy Instamart, jumped from about Rs 1,325 crore to roughly Rs 4,000 crore in 2025, a 202 percent surge, with projections near Rs 6,000 crore for 2026 (Inc42). Datum Intelligence puts total quick commerce ad spend at Rs 5,000 to 6,000 crore a year across categories (Storyboard18). That growth changes the bidding problem in one important way. On quick commerce, purchase decisions compress into the top few results in seconds, so share of those prime impressions tracks much closer to actual market share than a clean cost-per-click model would suggest. The maturity ladder still holds, but on these surfaces the early fixed-bid phase is less about cost discovery and more about buying enough visibility to learn whether the slot converts at all. If you are weighing where to put that budget first, our view on which quick commerce platform to launch on first is the sequencing call to make before you tune a single bid.

The short version

Dynamic bidding is not a smarter default. It is a tool that amplifies whatever signal you feed it, which makes it brilliant on proven keywords and wasteful on unproven ones. Fixed bids are how you learn cheaply. Dynamic bids are how you scale what you have learned. Rule-based bids are how you stay in control once the account gets big enough that manual management fails. With CPCs climbing and quick commerce ad budgets multiplying, the error is not choosing the wrong one. It is choosing one and standing still while your keywords mature past it.

Our Performance Marketing & Ads teams build accounts as layered structures, not single settings, so unproven terms learn on fixed bids while winners scale on dynamic, and our Analytics & Reporting work surfaces the conversion signal that tells you exactly when a keyword has earned the next rung. Pick the bid type that matches what you know. Then keep changing it as you learn more.

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.

Book a meeting →