Amazon vs Flipkart vs Myntra: where should a new seller start?

There is no defensible universal winner between Amazon, Flipkart and Myntra. Marketplace fees, eligibility, fulfilment choices, competition and return costs vary by product and change over time, so choose a first platform from current evidence for your own SKUs—not from a fixed ranking or an old fee table.

This guide is a seller decision framework. It does not assume that one platform has better buyers, lower returns or higher margins without account-level evidence.

Start with platform fit, not platform size

Amazon and Flipkart support broad product catalogues. Myntra describes its partner destination around fashion and beauty and applies defined selection criteria; its official partner portal says a qualifying brand can be onboarded within 15 business days. That is an eligibility statement, not a promise that every application will be accepted or completed in that time.

Use three filters before comparing fees:

1. Eligibility: Can this legal entity, brand and product category be accepted, and are all required documents available?
2. Demand evidence: Do comparable products show active recent reviews, stable availability and a range of credible sellers—not only sponsored placements or one dominant listing?
3. Execution fit: Can the team meet the platform's catalogue, dispatch, inventory and return workflows without weakening the rest of the business?

If a product fails one of these filters, a large marketplace audience does not make it the right first launch.

Compare the economics SKU by SKU

Do not compare headline commission percentages. Build the same contribution sheet for each shortlisted platform using its current terms and the fulfilment model you would actually use.

For each SKU, record:

• realised selling price after seller-funded discounts;
• product or landed cost;
• category-dependent marketplace charges;
• closing, collection or fixed charges where applicable;
• fulfilment, forward shipping, packaging and storage;
• return, RTO, damage and non-recoverable refund cost from your own comparable history;
• advertising allowance, if the launch needs paid discovery;
• taxes and credits reviewed on a consistent basis with a qualified adviser.

Amazon India's current fee page shows why a generic threshold is unreliable: referral fees vary by category, closing fees by price and fulfilment channel, and weight-handling fees by shipment characteristics. Amazon also labels its revenue-calculator outputs as estimates, so reconcile estimates with actual transaction and settlement data after launch.

For Flipkart and Myntra, use the rate card, agreement and transaction reports available to the specific seller account and category. Do not transfer an Amazon fee assumption into another platform or treat a bank payout as revenue; a payout can combine fees, refunds and adjustments from different orders.

Evaluate catalogue and fulfilment readiness

The first marketplace should be the one the team can operate cleanly. Amazon's seller overview presents multiple fulfilment routes, including Fulfilment by Amazon, Easy Ship and self-ship. Each route changes inventory placement, operational responsibility and cost, so model the route you will use rather than the platform name alone.

Before choosing Amazon or Flipkart, prepare:

• product identifiers and category attributes;
• compliant primary and supporting images;
• accurate dimensions, weight, tax and package data;
• titles, bullets and specifications that match the physical product;
• stock and dispatch ownership for every working day;
• a return-evidence and settlement-reconciliation process.

For Myntra, add brand-selection readiness, fashion or beauty catalogue depth, size and fit information where relevant, and the image and attribute assets required for the intended category. The official partner portal is the authoritative place to confirm current selection and onboarding instructions.

Use marketplace evidence without inventing demand

A search-results review is useful, but it is not keyword-volume research and it does not prove future sales. Run the same manual check on each platform for the exact product type and buyer wording.

Record:

• how many close substitutes appear in the first useful result set;
• their price and pack-size range;
• whether results are dominated by established brands or fragmented across sellers;
• review recency and the questions buyers repeatedly ask;
• delivery promises to representative serviceable pin codes;
• sponsored density and the likely need for launch advertising;
• visible gaps in images, sizing, compatibility, ingredients or specifications.

Keep screenshots or a dated observation sheet. Repeat the review before launch because marketplace results, prices and availability change. Do not publish a demand, market-share or buyer-quality claim unless the source and method support it.

Run a controlled first-platform test

Start with a small, representative SKU cohort instead of uploading the entire catalogue. Use products that are in stock, operationally simple and economically understandable. The test should answer whether the platform can acquire and fulfil orders while protecting contribution—not whether one launch week produced an exciting gross-sales number.

Define before launch:

• test SKUs and stock ceiling;
• price and promotion boundaries;
• fulfilment route;
• advertising cap and break-even guardrail;
• listing-quality checklist;
• observation window long enough to include settlement and return outcomes;
• stop, repair and expand rules.

Review impressions, clicks, conversion, cancellations, returns, RTO, advertising spend, net settlement and contribution together. A low order count can still expose catalogue rejection, poor click-through or fulfilment friction; it cannot establish a stable return rate or profitable scale.

Decide when to add the second marketplace

Expansion is justified when the first platform has a repeatable operating record, not merely because the seller account is live. Add the next marketplace when:

• the selected SKUs remain in stock reliably;
• catalogue errors and dispatch exceptions are controlled;
• actual settlements reconcile to the order-level model;
• returns are coded by reason and their cost is understood;
• advertising, where used, stays within a documented contribution limit;
• one owner can maintain platform-specific pricing, content and inventory.

Do not copy a listing mechanically. Preserve the product facts, but adapt attributes, image order and searchable wording to the destination platform's current catalogue structure. Keep separate performance and settlement ledgers so one channel's economics do not hide another's losses.

Make the choice with a dated scorecard

Score each shortlisted platform from 0 to 2 on five dimensions: eligibility, observed demand, contribution after variable costs, catalogue readiness and fulfilment capacity. Attach the evidence date and source to every score.

The scoring is a prioritisation tool, not a forecast. A platform with the highest evidence-backed score becomes the first test; a tie should be broken by the smallest safe operational test, not by an unsupported claim about which marketplace is universally best.

Recalculate after the first complete order, return and settlement cohort. The best first platform is the one that produces a repeatable process and defensible unit economics for the seller's actual products. The best long-term channel mix may be different once the business has evidence from more than one marketplace.