Two thousand units of liquid lipstick in four shades are stacked in your spare room, and the channel question is now a cash question. The standard D2C advice says build your own website and own your customer. In color cosmetics that advice needs surgery before you follow it. Makeup repeat is slower than skincare or haircare, 15 to 25% with a two-to-four-month gap between orders, so the own-site compounding that carries a shampoo brand arrives late here. And makeup conversion runs on proof of shade on real skin, photo reviews and swatches, which stack fastest on a marketplace listing. So here is the direct answer: in makeup, marketplaces deserve a bigger, earlier role than in any other beauty category. Nykaa, because it is where India shops color with intent. Amazon, because zero referral fees under ₹1,000 make validation cheap. But the kit, the cross-shade sale and the same-shade reorder, the three places makeup profit actually lives, only pay on your own site. You are not picking a side. You are assigning jobs to doors, in a sequence.
This guide prices all six doors at 2026 fees, runs one ₹499 liquid lipstick through the three that matter line by line, shows why a shade-mismatch return costs different money on different channels, and ends with a decision rule built on two numbers you already have: monthly orders and shade count. If you are a step earlier, still choosing the product and the shades, start with how to start a makeup brand in India and come back with stock in hand.
Makeup buyers buy a shade, not a product, and that rewrites every channel's economics. The 2026 fees: Amazon and Flipkart charge zero referral fee on products under ₹1,000, so a ₹499 lipstick pays only closing and fulfilment, but contested beauty keywords cost ₹100 to ₹140 an order in ads and the buyer stays masked. Nykaa, whose beauty GMV crossed ₹11,700 crore in FY25 with color roughly a third of it, takes 18 to 25% commission plus 18% GST on the commission, settles in 30 to 45 days, and gatekeeps onboarding. Quick commerce moves kajal and lipstick fast but keeps 30 to 35% all-in, with per-SKU listing fees that multiply by shade. Your own site keeps full margin, buys every visitor at a ₹130 to ₹200 CAC, and is the only door that can gate shade-mismatch returns before the order and run the kit, cross-shade and same-shade replenishment plays. First-order math favours marketplaces: roughly ₹125 on Amazon and ₹130 on Nykaa against ₹90 on your own site. But the ₹999 kit nets about ₹315 and the flow-triggered reorder about ₹280, and both exist only on your site. Validate where trust is rented. Retain where the customer is yours.
Why the channel question works differently in makeup
A shampoo buyer risks ₹449 on a promise. A lipstick buyer risks looking wrong in daylight, and she knows it. That one difference produces the four physics rules every fee table below sits on.
Conversion runs on shade proof. Nobody buys color from a render. They buy from a swatch on skin that looks like theirs, a photo review, a creator applying it on camera. That proof concentrates where reviews concentrate, which is a single marketplace listing, not your scattered Instagram comments.
Returns are shade returns, and the channel decides their cost. Makeup's unmanaged return and RTO band is 25 to 35%, the worst in beauty, because color disappointment stacks on COD refusal. Who eats that, and how often, depends on the door, which gets its own section below.
Repeat is slow and shade-loyal. A daily-wear kajal empties in six to ten weeks, a liquid lipstick in three to four months. Repeat runs 15 to 25%, against 30 to 45% in haircare, and when the reorder finally comes it is for the exact same shade. Loyalty in makeup attaches to a shade number, which cuts both ways: a one-tap reorder link wins it, a stockout in her shade loses her entirely, because nobody substitutes their brown.
Inventory multiplies by shade times channel. Every shade is its own SKU with its own MOQ and expiry clock. Every stocked channel, FBA, Nykaa's warehouse, your own shelf, is another position per shade. Four shades on three stocked channels is twelve inventory positions, each with a minimum. In the haircare version of this decision, the refill row settles the argument almost on its own. Makeup gives you no such shortcut, so the answer has to be built door by door.
The six doors: where makeup actually sells online in 2026
Six channels matter for a color cosmetics brand. Here is what each takes, what it really costs beyond the headline rate, and who is shopping there.
| Channel | Take per order | Other real costs | The makeup buyer there | Do you get the customer? |
|---|---|---|---|---|
| Your own store (Shopify or equivalent) | No commission; ~₹2,000/month platform + ~2% gateway (UPI 0% MDR) | Courier ₹35 to ₹85 by zone, RTO provision, Meta CAC ₹130 to ₹200 | Creator-led shade shoppers, and your entire reorder list | Yes: phone, email, shade history |
| Nykaa | 18 to 25% commission + 18% GST on the commission, 21 to 30% effective | Visibility ads, 30 to 45 day settlement, gated onboarding | The densest color-cosmetics intent in the country | No |
| Amazon | 0% referral under ₹1,000 (since March 2026); closing fee ~₹20 to 25 | FBA fulfilment ~₹70, 18% GST on fees, sponsored ads ₹100 to ₹140 | Claim searchers: smudge-proof kajal, transfer-proof lipstick; gifting spikes | No, contact details masked |
| Flipkart | 0% commission under ₹1,000; rate-card tiers above | Fulfilment + ads; thinner beauty review culture | Value-leaning, tier 2/3 heavy | No |
| Myntra Beauty | Tiered: entry promo near 2% to 15%+ standard, plus a Growth Enablement Fee | Gated 7 to 15 day onboarding, fashion-first merchandising | Premium, design-led beauty buyers, mostly women | No |
| Quick commerce (Blinkit, Zepto, Instamart) | 30 to 35% of MRP all-in | ~₹25,000 per SKU per state listing, and every shade is an SKU; ₹2 to 3 lakh/month marketing floor | The 7pm kajal emergency before an event | No |
Meesho is deliberately absent: its ₹99 to ₹199 price war sits below makeup's ₹299 to ₹699 band, and its buyers return more. Use it to clear a discontinued shade, never to build a positioned brand.
Nykaa: the beauty gravity well you earn into
Nykaa is not one channel among six for a makeup brand. It is the shelf your category lives on. Its beauty vertical closed FY25 at about ₹11,800 crore of GMV, growing 30% a year, on a base of 34 million beauty customers, and color cosmetics contributes roughly a third of it. The buyer arrives reading undertones and comparing swatches, not hunting the cheapest bottle, which is exactly why you cannot simply sign up. Onboarding is curated and the category team filters most small applicants out. Commission runs 18 to 25% by sub-category, with 18% GST charged on the commission itself, so the effective cut is 21 to 30% before visibility ads, and settlement lands 30 to 45 days after delivery, which means Nykaa revenue cannot fund next month's shade run the way your own store's T+2 settlement can. Many beauty brands operate on Nykaa's inventory model, where Nykaa buys stock at a negotiated wholesale margin, so build the math on the price Nykaa pays you, not your MRP, and remember the stock you park in its warehouse is per shade. Apply in month 6 to 9 with a trademark, per-shade batch COAs, rating proof and margin room to share. The application mechanics are in how to sell on Nykaa.
Amazon: cheap fees, brutal attention, gifting season
The fee side is the friendliest it has ever been. Since March 16, 2026, Amazon India charges zero referral fee on products under ₹1,000 across 1,800+ categories, beauty included, so a ₹499 lipstick pays a closing fee near ₹20 to 25 plus fulfilment and GST on those fees. Nearly all of makeup qualifies: singles at ₹299 to ₹699 and a kit held at ₹999 both sit inside the window, while a ₹1,199 kit pays the 5 to 18% category referral again, a real argument for pricing the marketplace kit at ₹999.
What stays expensive is attention. Search "matte liquid lipstick" and the page is a wall of labels from the same Mumbai, Thane and Baddi fillers, held by brands with thousands of photo reviews. A zero-review listing starts beneath all of them and pays ₹100 to ₹140 an order in sponsored ads for its first two quarters. The workable wedge is a claim term you can win and film, smudge-proof kajal for oily lids, transfer-proof bridal lipstick, not the head term. Amazon also owns the gifting spikes, Valentine's, Rakhi, Diwali, wedding season, when a gift-ready ₹999 kit moves without resistance. Treat it as a search-and-gifting harvester with a review wall you are deliberately building, not a home base. Listing method in how to sell on Amazon in India.
Flipkart and Myntra Beauty: the second listing and the premium shelf
Flipkart mirrors the zero-commission window under ₹1,000, so the fee case matches Amazon's. The demand case is thinner for color: less review depth, more value-hunting, a heavier tier 2/3 base where a ₹249 kajal moves better than a ₹599 lipstick. Switch it on as a low-effort second listing once Amazon runs itself, same photos, same copy.
Myntra Beauty is the shelf premium makeup founders under-price. Commission is tiered, entry promotional rates near 2% rising past 15% by category, plus a Growth Enablement Fee, with gated 7 to 15 day onboarding. The audience is fashion-first, female and premium, and the beauty push is real: Myntra's D2C beauty segment has grown at 105% year on year. The fit is design-led color that photographs well next to fashion. The wrong fit is a value kajal competing on price. Month 6+, with proof.
Quick commerce: real makeup demand, per-shade economics
Makeup genuinely fits the 10-minute apps, because the category's emergency is real: the kajal that ran out at 7pm before a dinner, the lipstick forgotten before a wedding. Beauty and personal care is already 13.4% of Blinkit's daily sales, and Zepto's beauty business has grown more than five times year on year, with eye makeup and lipsticks among its key drivers. Now the operator's read. The all-in take runs 30 to 35% of MRP once commission, storage and delivery are counted, the listing fee near ₹25,000 per SKU per state multiplies by every shade you list, and the marketing floor sits at ₹2 to 3 lakh a month. Dark stores also carry no ranges: your black kajal and two bestselling lip shades get shelf space, the other six shades never will. So quick commerce is a hero-shade channel for a brand with proven velocity, one or two cities, gates passed, never a launch door. Full mechanics in quick commerce for D2C brands.
Your own store: full margin, bought traffic, and the only shade gate
Your store keeps everything the platforms take: about ₹2,000 a month of platform cost, a gateway at roughly 2% plus GST with UPI at 0% MDR, settlement in T+2 to T+3 that funds the next shade run within the week, and a courier moving a lipstick parcel for ₹35 to ₹45 in-zone and ₹65 to ₹85 national. In exchange, nobody walks in. Every visitor is bought from Meta's most contested auction, the female beauty audience, at a ₹130 to ₹200 CAC for a new brand. What that buys you is the three jobs no marketplace will do: pages that lead with the ₹999 kit instead of the cheapest single, a shade-finder that gates mismatch returns before the order exists, and the customer's number with her shade attached. The platform-level trade is in Amazon vs Shopify in India.
The ₹499 test: one lipstick through three doors
Run one real SKU through the three main doors and each channel's personality shows up in the arithmetic. The product: a ₹499 liquid lipstick with ₹125 of landed cost (fill ₹80, pack ₹45), the 25% COGS setup from the makeup flagship.
| Line | Own site (Razorpay + Shiprocket) | Amazon (FBA) | Nykaa |
|---|---|---|---|
| Selling price | ₹499 | ₹499 | ₹499 |
| Product + packaging | −₹125 | −₹125 | −₹125 |
| Platform take | −₹12 gateway (~2% + GST; UPI orders free) | −₹35 (₹0 referral + ~₹20 to 25 closing + GST on fees) | −₹118 (20% commission + 18% GST on it) |
| Fulfilment | −₹65 courier, blended zones | −₹70 FBA pick, pack, ship | −₹60 partner logistics |
| Returns / RTO provision | −₹45 (15% on a COD-mixed funnel, shade-gated) | −₹25 (COD refusals + claims) | −₹10 (prepaid-leaning, high intent) |
| Cost of demand | −₹160 Meta + creator CAC | −₹120 sponsored ads | −₹55 visibility ads |
| Net, first order | ~₹90 | ~₹125 | ~₹130 |
| Net, same-shade reorder | ~₹280 (flow-triggered, prepaid) | ~₹125, if she finds you again | ~₹130, if she finds you again |
Margin Waterfall™: selling price minus COGS, packaging, fulfilment, platform fees, RTO loss, then CAC or ad cost, run separately for every channel on the table. In makeup, run it three times per channel: on the single at first-order cost, on the same-shade reorder at that channel's real reorder cost, which is ₹10 to 20 for a WhatsApp flow on your own list and a fresh ad auction everywhere else, and on the ₹999 kit. The first pass shows where validation is cheap. The second shows who owns the shade-loyal customer. The third shows what to lead with.
Read it honestly, because makeup's read differs from every other beauty category. On the first order, the marketplaces win: ₹125 to ₹130 against your site's ₹90, because their beauty buyer costs less to reach than a cold Meta audience. In haircare you can wave that away, the refill arrives in six weeks and flips the table. In makeup the reorder takes two to four months and only 15 to 25% of buyers make it, so the marketplace's first-order edge is worth more here, and pretending otherwise burns cash. What flips the table in makeup is not the reorder alone. It is the cart. Lead with the ₹999 kit, a lip duo plus kajal at ₹335 landed cost, and your own site nets about ₹315 on the first order, more than double the marketplace single, because courier and gateway barely move while the cart doubles. Add the reorder row, ₹280 flow-triggered and prepaid against ₹125 in a fresh auction, and the own-site column compounds from order one. Marketplaces win singles. Your site wins carts and the second order. Build the plan on that split.
Shade-mismatch returns: the cost that changes by door
The shade looked different on screen. That sentence drives makeup's 25 to 35% unmanaged return and RTO band, and each round trip costs ₹120 to ₹250 in forward plus reverse shipping before repacking and loss. On a lipstick netting ₹90, one RTO erases two to three delivered orders. What most founders miss: the same disappointed buyer costs you differently on each channel.
On Amazon and Flipkart, most cosmetics are listed non-returnable for hygiene, but the policy carves out damaged, defective or wrong-item claims, which still get refunded or replaced, and platforms adjudicate those buyer-first, so "shade not as shown" often lands as a claim you fund. The bigger leak is the doorstep: a COD buyer whose enthusiasm faded in the four days since ordering refuses the parcel, and you cannot call her first, because she was never your customer. You pay both legs of the trip.
On Nykaa, the mismatch is structurally lowest: buyers are prepaid-leaning, arrive knowing their undertone, and shop from swatch-dense pages. Part of that 21 to 30% commission is, in effect, a return-prevention fee.
On your own site, and only there, you can gate the mismatch before the order exists: every shade shot on three or four real skin tones in daylight, an undertone guide in plain words, a three-question shade finder before the buy button, and a WhatsApp confirmation on every COD order. That stack is why a disciplined makeup brand holds prepaid past 55% and RTO near 12 to 15% while COD-heavy range sellers bleed at 30%. The full playbook is in how to reduce RTO on COD orders. Marketplaces price returns into your margin. Your site lets you prevent them, and prevention is the cheaper business.
Swatch content and review velocity: the conversion engine everywhere
Whatever mix of doors you run, the thing that converts is identical: shade on a real face, in motion, on skin the buyer recognises. A creator applying your brown in daylight outsells a studio render on every channel at once, which is why the swatch stack is the one asset you build before any channel goes live. Shoot it once, properly: every shade on three or four skin tones, photo plus 15-second video, and feed the same truth to your product pages, your ads, your Amazon image stack and your Nykaa listing. The system for sourcing that content at small-brand budgets is in the UGC playbook for D2C.
Reviews follow the same physics. Expect one to three organic reviews per 100 delivered orders. On a marketplace they concentrate on one listing where rank and conversion compound: 300 orders a month builds a 15 to 25 photo-review wall in a quarter, and in makeup a photo review is a swatch, the most persuasive object in the category. On your own site the same proof scatters across Google and Instagram comments. So rent the velocity deliberately: the Amazon listing is where your proof wall rises fastest, and those rating screenshots later become exactly the traction evidence a Nykaa category manager asks for. Just bank the relationship while you rent the shelf: a ₹3 insert in every parcel, marketplace parcels included, with a shade-care card, a QR to your shade finder and a first-reorder code. That insert is the bridge from rented proof to owned customers.
Repeat, shade loyalty, and why the sequence differs from haircare
Makeup's repeat curve is the slowest in beauty: 15 to 25%, on a six-to-ten-week cycle for kajal and three to four months for lipstick. Two operating consequences follow.
First, marketplace economics matter more, longer. A haircare brand can subsidise a thin first order because the refill lands in 45 days. A makeup brand waiting three months for a maybe-reorder cannot, so the doors that deliver profitable first orders, Amazon inside the zero-referral window, Nykaa once you are in, carry real weight in years one and two, not just launch month.
Second, when the reorder comes, it is winner-take-all on availability. Shade loyalty means she reorders 04 Brick, not "a lipstick". If your site holds her number and her shade, a day-90 nudge with a one-tap link wins at near-zero cost. If she reorders through Amazon search, you pay the auction again while three lookalikes bid above you. And if your site is out of her shade, she does not substitute, she leaves. That is why the own-site retention stack in makeup is timed by product and pinned to shade: a day-60 WhatsApp nudge for kajal, day-90 for lipstick, back-in-stock alerts by shade, the kit upsell on the second order, and the cross-shade drop, if she wears 04 Brick, show her the terracotta from the same family, not the whole card. Flow templates and send rules are in WhatsApp marketing for D2C. Cross-shade is the quiet engine here: the second shade to a proven buyer carries no CAC, no mismatch risk she has not already accepted, and it is how a 15 to 25% repeat rate still compounds into haircare-grade customer value.
At Atomberg, through the ₹400cr to ₹1,200cr stretch, the discipline that saved us in every channel review was counting inventory positions, not channels. Every new door multiplied SKU-level stock points, and the tail of slow movers quietly ate the working capital the winners needed. Makeup is that problem at its sharpest, because every shade is a full SKU with its own MOQ and its own expiry clock. Before any founder adds a channel, I make them write one multiplication: live shades times stocked channels. Four shades on three doors is twelve positions, each needing a minimum, and the hero shade will starve on the channel that is selling while the losers age in a warehouse you cannot see. If the multiplication scares you, the answer is almost never fewer channels. It is fewer shades, stocked deeper, and a channel plan that shades earn one at a time.
The decision rule: orders and shade count pick your channels
Positioning tells you where to start. Two numbers tell you when each next door earns its overhead: last month's orders, and how many shades are live. Volume decides whether a channel's fixed work pays back. Shade count decides whether you can stock it without starving your winner.
Under 150 orders a month → one engine only. A claim-led, shade-light product (black kajal, one perfect brown) validates on Amazon inside the zero-referral window; an undertone-wedge brand at ₹399+ with a creator engine starts on its own site; pre-launch, run the ₹50,000 one-shade play first. Shade gate before any expansion: never open a second stocked channel until every live shade can stay in stock on both, and if you carry more than 4 shades on under 300 orders a month, trim the range before adding a door. 150 to 500 orders → own site + Amazon both live, Flipkart on with the same assets, WhatsApp flows running, prepaid past 55%. 500 to 1,000 with a photo-review wall and a margin sheet that survives a 21 to 30% effective cut plus 30 to 45 day settlement → start the Nykaa file; open the Myntra Beauty conversation if the brand is premium and design-led. 1,000+ with 20%+ repeat, 65%+ gross margin and two or three hero shades carrying most volume → quick-commerce pilot, hero shades only, one or two cities, listing fees budgeted per shade. At any volume, if repeat sits under 10% and kit share under 20% of orders → stop adding doors and fix the retention stack first.
The hybrid sequence a makeup brand should run
According to the Founder Decision Loop™, channel expansion follows proof, not ambition: each door opens when the previous one produces the data that justifies it. For a positioned makeup brand, the calendar runs like this.
- Month 0 to 1: own store live, kit at ₹999 leading every page, shade finder before the buy button, swatch stack shot on real skin tones, WhatsApp opt-in at checkout, prepaid pushed past 55%.
- Month 1 to 2: Amazon live on one winnable claim term, earlier than you would in haircare, because the review wall takes two quarters to compound and zero referral makes the wait cheap. Insert card in every parcel. One MRP everywhere.
- Month 3 to 4: Flipkart switched on with the same assets. Day-60 kajal and day-90 lipstick flows start firing. First cross-shade drop to buyers past their second order.
- Month 6 to 9: Nykaa application with trademark, per-shade batch COAs, Legal Metrology labels, rating screenshots and a margin sheet with room to share; Myntra Beauty file if the brand photographs premium.
- Month 9 to 12: quick-commerce pilot with the hero shades in one or two cities, only with the volume, repeat and margin gates passed.
The public arcs back the split. Sugar built D2C-first on its own app and site, scaled to ₹505 crore in FY24 with a ₹76 crore loss, proof that owning the customer still means paying heavily for attention at scale. Swiss Beauty went the other way, marketplace and retail-first into 25,000 touchpoints and around ₹425 crore, bootstrapped. Both ended hybrid. Neither stayed on one door, and neither will you.
Splitting a four-shade range across three stocked channels before any shade earned depth. A founder with 4,000 units, four shades at 1,000 each, roughly ₹5 lakh of landed inventory, sends 300 per shade to FBA, 300 per shade to Nykaa's warehouse, and keeps the rest for the website. Twelve inventory positions. By festive week the data is in: one shade is 55% of demand, and it sells out on Amazon, the channel that was winning, mid-sale. Restocking means a fresh 1,000-unit per-shade run on a four-to-five-week lead time, so the listing's rank decays exactly when it was compounding. Meanwhile 1,600 units of the two slow shades age in three warehouses toward the 75%-remaining shelf-life rule marketplaces enforce at inward, come home unsellable through serious channels, and clear at 60% off. Cost: about ₹1 to 1.2 lakh in liquidation plus the festive rank that never came back. Depth on winners in one channel beats width everywhere. Shades earn channels one at a time.
Execution checklist
- Write last month's orders and your live shade count at the top of one sheet; that pair picks your channels, not a funded brand's story.
- Run the Margin Waterfall™ on your hero shade for every door, first order and same-shade reorder separately, then on the ₹999 kit.
- Shoot the swatch stack once, properly: every shade on 3 to 4 real skin tones in daylight, photo plus video, and feed every channel from it.
- Own store: kit-led pages, shade finder before the buy button, WhatsApp opt-in at checkout, back-in-stock alerts by shade.
- Amazon: confirm the single and the kit sit under ₹1,000 on the live rate card, hold the kit at ₹999, and target one winnable claim term, never "matte lipstick".
- Print the ₹3 insert, shade-care card, QR to the shade finder, first-reorder code, and pack it in every parcel on every channel.
- One MRP everywhere; offers run as own-site kits and reorder codes, never as marketplace price cuts on the hero shade.
- Time flows by product: day-60 kajal nudge, day-90 lipstick nudge, cross-shade drop only after the second delivered order.
- Nykaa file at month 4 to 6: trademark, per-shade batch COAs, Legal Metrology labels, rating screenshots, and a margin sheet that survives a 21 to 30% cut plus 30 to 45 day settlement.
- Quick commerce only past 1,000 orders a month, 20%+ repeat and 65%+ gross margin: hero shades only, one or two cities, listing fees budgeted per shade per state.
Your next action
Tonight, one sheet. Three columns: own site, Amazon, Nykaa. Two rows: what your hero shade nets on a first order and on a same-shade reorder, using your real fill cost, your real courier slab and the fee numbers above. Then write two numbers at the top, last month's orders and your live shade count, and multiply the shade count by the channels you were about to stock. The reorder row tells you why the list always ends at your own site. The multiplication tells you whether your inventory can survive the plan. Thirty minutes of arithmetic settles a debate most founders run on instinct for a year.
If you'd like the complete execution system, calculators, SOPs, templates and operating frameworks behind this process, continue inside D2C Acquisition.Lab.
