Your first 500 bottles are boxed, the labels are on, and the channel argument starts. Own site or Amazon. In every other category that argument is about fees and customer data. In perfume it starts with a truck. Alcohol-based fragrance is a flammable liquid, most Indian couriers will not put it on a plane, and surface-only movement is your base case rather than an exception. That one fact moves your delivery promise, your serviceable pin codes and your cost per parcel before anybody opens a rate card.
So here is the direct answer. Your own site is home base from day zero, because it is the only door where you control the delivery promise on a surface parcel and the only place the discovery set funnel works. Amazon comes second, month 2 to 3, and it has to be priced honestly, because fragrance did not get the zero referral fee under ₹1,000 that most categories got in March 2026. A ₹799 Eau de Parfum pays 14%. A ₹499 lipstick on the same platform pays nothing. Nykaa is a month 6 to 9 decision that earns its 21 to 30% effective cut. Flipkart is cheap incremental revenue. Myntra only if the bottle photographs premium. Fees rank the doors. The parcel decides which ones you can actually serve.
Perfume's channel decision is settled by logistics first and fees second. Alcohol-based EDP is a Class 3 flammable liquid, restricted for air movement, so it travels surface: longer SLAs, a smaller serviceable pin-code map, dangerous goods handling and a partner list that shrinks to whoever will sign for it. On fees, the surprise is Amazon. Beauty is not one rate card. The fragrance row charges 0% up to ₹500, 14% from ₹501 to ₹1,000 and 10% above ₹1,000, so a ₹799 bottle pays the highest referral rate anywhere in beauty on that platform. Run one ₹799 EDP with ₹190 of landed cost through five doors and contribution per kept order lands near ₹251 on your own site, ₹205 on Amazon, ₹276 on Flipkart, ₹261 on Nykaa and ₹280 on Myntra. Three of the four marketplaces win the first order. Amazon does not, because the closing fee in that band is ₹45 on top of the 14%. Your site wins everything after it, because the repeat bottle nets about ₹476 with no acquisition cost attached, and because it is the only channel that can tell a buyer the truth about the delivery date. Sequence it: own site from day zero, Amazon month 2 to 3, Flipkart month 3 to 4, Nykaa month 6 to 9, Myntra only when the brand has earned that shelf.
The shipping wall: why perfume's channel plan starts with the parcel
An Eau de Parfum is mostly ethanol. That makes it a Class 3 flammable liquid, shipped under UN 1266 for perfumery products, and in Indian air movement it sits under the Aircraft (Carriage of Dangerous Goods) Rules, 2003, which allow dangerous goods on aircraft only through operators certified to carry them and only against signed dangerous goods documentation. No express network wants that paperwork attached to a ₹799 retail parcel. Blue Dart bans all IATA restricted items and dangerous goods across every one of its services and lists liquids and semi-liquids as banned with conditions. Aggregator panels behave the same way in practice: some partners refuse fragrance outright, some move it surface-only, a few accept it with a declaration and a surcharge.
Four consequences follow, and every one of them is a channel decision, not a warehouse decision.
- Your SLA is surface, so write surface. A metro-to-metro air parcel lands in 1 to 2 days. The same parcel on surface takes 3 to 5 days in-zone and 5 to 8 days national, longer into the North East, Kashmir and the islands. Every channel that lets you set the promise has to be set to the truth, and the ones that set it for you have to be watched.
- Your pin-code map shrinks. Your serviceable list is not your aggregator's list. It is the overlap between your aggregator's list and the partners who accept fragrance. Pull both, overlay them on last month's orders, and mark the gap before you switch on all-India delivery. The allocation logic is in multi-courier allocation strategy, and the panel-by-panel comparison in Shiprocket vs NimbusPost vs Delhivery.
- Your per-parcel cost carries a premium. Budget above your normal 0.5 kg slab for handling, declarations and a thinner partner set. Get the exact surcharge in writing from your aggregator before you model anything, because it is quoted per account and it moves.
- Marketplace fulfilment programmes gate you. Amazon runs a dangerous goods review on any listing suspected of hazmat content and asks sellers for a safety data sheet. Plan for FBA to be limited or refused and for Easy Ship or self-ship to carry the volume. Ask your filler for the SDS on day one, not in the week you list.
If you are still choosing the juice and the filler, the upstream decisions sit in how to start a perfume brand in India. Come back here with stock in hand and a courier list in writing.
The five doors, priced on a ₹799 bottle
| Door | Take on a ₹799 EDP | Other real costs | The dangerous goods reality | Do you own the customer? |
|---|---|---|---|---|
| Your own store | No commission. Gateway about 2% plus GST on card orders, UPI at 0% MDR | Platform near ₹2,000 a month, courier ₹95 blended on a surface DG parcel, Meta CAC ₹150 to ₹220 | You pick the partners and you write the promise. The only door where you can refuse a pin code before the order exists | Yes. Phone, email, scent history |
| Amazon | 14% referral in the ₹501 to ₹1,000 band, so ₹111.86, plus a ₹45 closing fee on Easy Ship, plus 18% GST on both | Weight handling on Easy Ship or FBA, sponsored ads ₹90 to ₹140 an order | Dangerous goods review before you list, SDS required, FBA often limited | No |
| Flipkart | 0% commission under ₹1,000. Fixed fee near ₹40 and a collection fee near ₹30, plus 18% GST on both | Flipkart shipping fee by weight and zone, thinner fragrance demand, heavier COD base | Same restricted-item logic through its logistics arm. Confirm fragrance acceptance for your seller account | No |
| Nykaa | 18 to 25% commission plus 18% GST on the commission, so 21 to 30% effective | Visibility ads, 30 to 45 day settlement, gated onboarding, stock parked per SKU | Its network already handles beauty including fragrance, which removes a problem you would otherwise own | No |
| Myntra | Tiered, roughly 2% promotional entry to 15% and above standard, plus a Growth Enablement Fee and 18% GST | Gated 7 to 15 day onboarding, fashion-first merchandising | Fashion logistics with beauty bolted on. Confirm fragrance handling in the seller agreement | No |
Meesho is deliberately not a row here. Its zero commission is real, but its buyers price fragrance at ₹149 to ₹299, which works only for a deliberate value line in cheap stock bottles. Nothing you positioned at ₹799 belongs on that shelf.
The Amazon fragrance fee almost nobody checks
In March 2026 Amazon India expanded zero referral fees to over 12.5 crore products priced under ₹1,000 across 1,800+ categories. Most founders read that headline, assume beauty is covered end to end, and model a ₹799 bottle at zero. Beauty is not one rate card. Pull the live schedule and the fragrance row reads differently: Beauty · Fragrance charges 0% up to ₹500, 14% above ₹500 and up to ₹1,000, and 10% above ₹1,000, while Beauty · Makeup gets the full 0% under ₹1,000. Same platform, same department, about ₹112 of difference on a single order.
Now read that band structure as a pricing sheet, because it has a hole in it. At ₹999 the referral fee is ₹139.86. At ₹1,049 the rate drops to 10% and the fee is ₹104.90. The closing fee moves at the same line too, ₹45 up to ₹1,000 and ₹76 above it on Easy Ship, so count both: ₹999 nets ₹814 after fees and ₹1,049 nets ₹868. You raise the price by 5% and keep ₹54 more on every order. There is no price inside the 14% band that beats it. If your Amazon hero is sitting at ₹999 because ₹999 feels right, that number is costing you twice: once on the rate and once on the rupees.
The mirror image is just as useful. A ₹399 or ₹499 discovery set pays a 0% referral, because it sits under the ₹500 line. The product that solves your biggest objection is also the cheapest thing you can sell on Amazon. On top of the referral sits a closing fee, and in the ₹501 to ₹1,000 band it is not small: ₹27 on FBA, ₹35 on Seller Flex, ₹45 on Easy Ship and ₹51 on self-ship, and every one of those goes up ₹3 from 7 September 2026. Then weight handling, where Easy Ship's standard slabs are ₹55 for the first 500 g, ₹75 from 500 g to 1 kg and ₹112 from 1 kg to 2 kg. A boxed 50 ml bottle with real protective packing lands at 400 to 500 g, so tight packing pays ₹55 and a lazy carton pays ₹75. Confirm every one of these on your own live fee page before you price, because this card moves. Listing mechanics are in how to sell on Amazon in India.
The ₹799 test: one bottle through five doors
Every number below runs on one blended basket, stated so you can rebuild it with your own figures. One 50 ml Eau de Parfum, MRP ₹799 including GST, one bottle per order, no gift sets and no discovery sets mixed in. Landed cost ₹190: juice ₹52, bottle with atomizer and cap ₹88, carton with label and shrink ₹32, filling, capping and QC ₹18. That is 24% of MRP, inside the 15 to 30% band this category runs on, and look at where it sits. The fragrance is ₹52. The glass, the pump and the box are ₹120. Founders who spend three months negotiating compound price are optimising a quarter of the bill. Own-site payment mix is 55% prepaid and 45% COD. Marketplace mixes are whatever each platform delivers.
| Line | Own site | Amazon | Flipkart | Nykaa | Myntra |
|---|---|---|---|---|---|
| Collected price | ₹799 | ₹799 | ₹799 | ₹799 | ₹799 |
| Product and packaging | −₹190 | −₹190 | −₹190 | −₹190 | −₹190 |
| Fulfilment (surface DG parcel) | −₹95 | −₹80 | −₹95 | −₹75 | −₹80 |
| Platform take (fees + 18% GST on fees) | ₹0 | −₹185 | −₹83 | −₹207 | −₹171 |
| Payment and collection | −₹19 | in fees | in fees | in fees | in fees |
| Cost of demand | −₹185 | −₹110 | −₹90 | −₹55 | −₹60 |
| Contribution on a clean order | ₹310 | ₹234 | ₹341 | ₹272 | ₹298 |
| RTO and refusal drag per kept order | −₹59 | −₹29 | −₹65 | −₹11 | −₹18 |
| Net per kept order | ₹251 | ₹205 | ₹276 | ₹261 | ₹280 |
| Net on the next bottle | ₹476 (day-45 flow, prepaid) | ₹205, if she finds you | ₹276, if she finds you | ₹261, if she finds you | ₹280, if she finds you |
The fee arithmetic behind those platform lines, so you can check it: Amazon is 14% of ₹799, which is ₹111.86, plus the ₹45 Easy Ship closing fee that applies from ₹501 to ₹1,000, and 18% GST rides on both, so ₹185. Nykaa at a mid-band 22% is ₹175.78 plus 18% GST, so ₹207. Myntra at 15% is ₹119.85 plus a Growth Enablement Fee near ₹25, plus GST on both, so ₹171. Flipkart takes no commission under ₹1,000, so it is a ₹40 fixed fee and a ₹30 collection fee plus GST, so ₹83. Own site pays ₹19 of payment cost: across 100 orders, 45 COD parcels at a ₹35 collection fee is ₹1,575, and of the 55 prepaid roughly 40 arrive on UPI at 0% MDR while 15 come on cards at about 2% plus GST, which is ₹283. That is ₹1,858 over 100 orders.
Read the table the way an operator would, not the way a founder forum would. On the first order the marketplaces win, and it is not close: ₹280 on Myntra, ₹276 on Flipkart, ₹261 on Nykaa against ₹251 on your own site. You are paying ₹185 to buy a stranger in the most crowded gifting auction in the country, and no fee saving covers that gap. Flipkart's zero commission under ₹1,000 makes it the cheapest rate card in the set, which is why founders who dismiss it on vibes leave money on the table for two quarters. Then look at the last row. Your site's second bottle nets ₹476, because reaching that customer costs about ₹8 of messaging and the checkout is prepaid, so the refusal risk nearly disappears. On every marketplace the second bottle nets exactly what the first one did, and only if she stumbles back into a search you are still paying for. That is the whole argument. Marketplaces rent you demand at a fair price. Your site is the only place a signature scent turns into a repeat without being called a subscription. Flow timing and templates are in WhatsApp marketing for D2C.
Margin Waterfall™: selling price minus COGS, packaging, fulfilment, platform fees, RTO loss, then CAC, run separately for every door on the table. In perfume, run it three times per channel: on the first bottle, on the repeat bottle at that channel's real reorder cost, and on the discovery set. The first pass tells you where validation is cheap. The second tells you who actually owns the customer. The third tells you what to lead with, which in this category is almost never the bottle.
Where GST actually sits in this table
Your ₹799 contains ₹122 of output GST at the 18% perfume slab, so your working revenue is ₹677. The tables above are drawn on the collected price because that is how a settlement report reads, but do the GST arithmetic once so you know what you are looking at. The ₹185 Amazon platform take contains about ₹28 of GST, and for a registered seller that ₹28 is input credit, not a cost. So is the ₹29 inside your ₹190 landed cost and the ₹14 inside your ₹95 freight bill. Net GST leaving you is about ₹78 on an own-site order and about ₹52 on Amazon, because the higher-fee channel throws off more credit. The 18% on marketplace fees is a cash-flow cost and a filing discipline, not a margin cost, provided you claim it. Marketplaces also withhold TCS under Section 52 of the CGST Act, 0.5% of net taxable supplies since 10 July 2024, which is about ₹3.40 on this order. It lands in your electronic cash ledger, so it is a timing cost, not a loss.
Non-returnable is not the same as no risk
Perfume is largely listed non-returnable on marketplaces once the seal is broken, for hygiene reasons, and founders read that as a gift. It is not. Amazon's returns policy still refunds or replaces damaged, defective or wrong-item deliveries even on non-returnable products, and platforms decide those buyer-first. So the risk does not vanish. It moves into three places, and none of them behaves like an apparel return.
COD refusal at the door. This becomes your main failure mode. A buyer who ordered on Monday and is handed a bill on the following Monday, because the parcel moved by road, has cooled off. On a marketplace you cannot call her first, because she was never yours. You pay both legs. The prepaid conversion playbook is in COD vs prepaid strategy.
Glass on a surface network. Roughly 11% of unit loads arrive damaged in Indian ecommerce, and a longer surface route adds handling touches and days to that exposure. A leaked bottle is a write-off, not a restock, and it usually spoils whatever is packed next to it. Spend ₹8 to ₹12 on a moulded insert and a shipper that survives a drop test before you spend it on a heavier cap. The packing spec is in ecommerce shipping and packaging protection.
The review you cannot refund. This one costs the most and appears in no report. In apparel a disappointed buyer sends the product back. In perfume she cannot, so she writes the review instead. "Smells nothing like the description" and "gone in an hour" are permanent assets belonging to your competitor, and on a marketplace that review is your conversion rate. The only real defence is the discovery set. Sell 2 ml vials before you sell the bottle, and the person who eventually pays ₹799 has already smelled it.
Now put a number on all three, using the model every budget page on this site runs on. RTO drag per delivered order equals the failure rate divided by one minus the failure rate, multiplied by the full cost of a failure. At a 12% refusal rate that is 0.136 failed parcels for every order you keep, not 0.12, which is where most founders understate the damage by about 14%. The failure itself costs ₹435 on your own site: ₹95 forward, ₹95 reverse, ₹32 of destroyed outer packaging, ₹185 of burnt acquisition spend and a ₹28 allowance for glass that comes back unsellable. That is ₹59 on every order you keep, failed or not. On Nykaa the same maths gives ₹11, because a prepaid, high-intent buyer barely refuses, and that ₹48 gap pays back nearly a quarter of Nykaa's commission on its own. The reduction playbook is in how to reduce RTO on COD orders.
Per order and per kept order, side by side
| Channel | Fees + fulfilment as % of ₹799 | Cost of demand | RTO drag per kept order | Contribution, clean order | Net per kept order | Net on the next bottle |
|---|---|---|---|---|---|---|
| Own site | 14% | ₹185 | ₹59 | ₹310 | ₹251 | ₹476 |
| Amazon | 33% | ₹110 | ₹29 | ₹234 | ₹205 | ₹205, unprompted |
| Flipkart | 22% | ₹90 | ₹65 | ₹341 | ₹276 | ₹276, unprompted |
| Nykaa | 35% | ₹55 | ₹11 | ₹272 | ₹261 | ₹261, unprompted |
| Myntra | 31% | ₹60 | ₹18 | ₹298 | ₹280 | ₹280, unprompted |
Two columns decide the plan. The kept-order column says start where demand is cheap. The next-bottle column says finish where the customer is yours. A brand that reads only the first column stays a marketplace supplier forever at ₹205 to ₹280 a sale. A brand that reads only the second column burns ₹185 a stranger for a year and runs out of cash before the repeat curve arrives. The perfume answer is both, in sequence, with the discovery set doing the heavy lifting at the top of every funnel. Flipkart's mechanics are in how to sell on Flipkart and Nykaa's gate is in how to sell on Nykaa.
What the discovery set does to this table
Price a 5-vial set at ₹399 and three things change at once. It pays a 0% referral on Amazon, because it sits under ₹500. It ships lighter, which matters on a network that charges by slab. And it converts far better than a cold ₹799 bottle, because the buyer is risking the price of two coffees on something she cannot judge from a photo. Run it: ₹399 collected, ₹75 landed for vials, juice and card box, ₹75 freight, ₹12 payment, ₹95 acquisition, and a ₹31 refusal drag at a 10% rate leaves about ₹111. Thin. But if 30% of those buyers come back for a full bottle worth ₹476 of contribution, each discovery buyer is worth roughly ₹254, slightly ahead of the ₹251 cold bottle, on a funnel two to three times wider at the top. The set is not a margin play. It is how you buy customers in a category where nobody can smell the product before paying.
In my distribution years the map that decided the month was never the sales map, it was the serviceability map. Perfume founders skip it because ecommerce taught everyone that every pin code is the same. It is not, and on a dangerous goods parcel it really is not. The first thing I make a fragrance founder do is pull the aggregator's fragrance-approved partner list, overlay it on last month's orders, and mark every pin code where the promise and the network disagree. It is usually 10 to 15% of orders, and almost every refusal lives inside that gap. Then comes the uncomfortable part: publish the honest date. A page that says seven days and delivers in six converts better than one that says two and delivers in six, because the second one manufactures a refusal at the door and a one-star review right behind it.
The sequence that actually works
According to the Founder Decision Loop™, a channel opens when the previous one has produced the proof that justifies it, not when a competitor announces a launch. For perfume the gates are volume, delivery reliability and margin room, in that order.
Under 100 orders a month → one door only, your own site, with a written fragrance acceptance list from your aggregator and an honest zone-wise SLA on every page. Lead with the discovery set, not the bottle. 100 to 300 orders → add Amazon, with the fragrance rate card already in your pricing sheet and the dangerous goods review cleared before inventory ships. If your hero would land at ₹999, price it at ₹1,049 and take the 10% band. 300 to 600 → switch on Flipkart with the same assets, because zero commission under ₹1,000 is the cheapest incremental revenue available to you. 600 to 1,000 with a review wall and a margin sheet that survives a 21 to 30% cut plus 30 to 45 day settlement → open the Nykaa file. 1,000+ with a bottle that photographs premium → start the Myntra conversation. At any volume, if on-time delivery on national pin codes sits under 85%, stop adding doors and fix the courier mix first, because every new door multiplies the same broken promise.
- Month 0. Aggregator's fragrance-approved partner list in writing. SDS and the filler's CDSCO licence copy in your folder. Zone-wise delivery dates on every product page. Discovery set at ₹399 leading the funnel. WhatsApp opt-in at checkout, prepaid pushed past 55%.
- Month 2 to 3. Amazon. Dangerous goods review cleared before inventory moves, referral modelled at 14% or the hero priced above ₹1,000, and one winnable scent-family term rather than a head term like perfume for men.
- Month 3 to 4. Flipkart with the same photos and copy. The day-45 refill nudge goes live on your own list.
- Month 6 to 9. Nykaa file: trademark, the filler's licence copy, batch COAs, Legal Metrology labels, rating screenshots and a margin sheet with room to share.
- Month 9 to 12. Myntra if the bottle photographs premium. Quick commerce stays a later question, because per-SKU listing fees and the storage side of flammable stock both need answers before it is worth modelling.
Copying a skincare brand's delivery promise onto a fragrance store. A founder went into Diwali week with "2 to 3 day delivery" on every product page, because that is what his theme shipped with, and pushed 600 parcels all-India. Most of them moved surface at 5 to 8 days, because the air partners in his panel bounced fragrance and the aggregator quietly reallocated. COD refusal ran at 22% instead of the 12% his sheet assumed. That is 60 extra failed parcels at about ₹435 each of forward, reverse, packaging, burnt ad spend and broken glass. ₹26,100 gone in one week, in the one week of the year that funds the quarter, plus a late-delivery rating hit that held his listing down until January. The fix cost nothing: pull the fragrance-approved partner list, publish zone-wise dates, and promise seven days wherever seven days is the truth.
Execution checklist
- Get the fragrance-approved courier list from your aggregator in writing, with the surface SLA, the handling surcharge and the pin-code exclusions, before you promise a date on any channel.
- Ask your filler for the safety data sheet and a copy of the CDSCO cosmetic manufacturing licence on day one. Both get asked for later, always at the worst moment.
- Open the live Amazon fee schedule and read the Beauty · Fragrance row yourself. Price the hero under ₹500 or above ₹1,000, never at ₹999.
- Rebuild the five-column table with your landed cost, your courier slab and your real CAC. Two rows matter: contribution on a clean order and net per kept order.
- Compute the RTO drag properly: rate divided by one minus rate, times forward plus reverse plus packaging plus burnt CAC plus a glass write-off allowance.
- Publish zone-wise delivery dates on your product pages and match them in every marketplace listing. Under-promise on purpose.
- Put the ₹399 discovery set at the top of every funnel, on every door, with the full bottle behind it.
- Print a ₹3 insert for every parcel on every channel: scent notes card, QR to your refill page, first-reorder code. It is the only bridge off a rented shelf.
- Set the day-45 refill nudge live before you open the second channel, because the repeat bottle is where this entire model pays.
Your next action
Two calls today, one sheet tonight. Call one goes to your aggregator: ask in writing which courier partners accept alcohol-based perfume, on which service, at what surcharge, and to which pin codes. Call two goes to your filler for the safety data sheet and the licence copy. Then open the Amazon fee page, find the Beauty · Fragrance row, and rebuild the five-column table above with your own landed cost, your own courier slab and your own CAC. If your hero sits at ₹999, move it tonight. Two phone calls and thirty minutes of arithmetic settle a channel debate most perfume 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.
