Marketplaces and quick commerce

What the sellers who did well changed

None of the three stories below is about AI. They are here because they show which decisions matter on these channels, and those decisions are made by people.

A kit for one occasion. Moms Home sold curated newborn and hospital-bag kits, each built around one moment in a customer's life [1].

A pack built for the channel. Pee Safe overhauled its supply chain in late 2023 and early 2024, grew its quick commerce range from roughly 5 to 6 SKUs to more than 60, and changed pack sizes. A 12-piece acne patch pack at ₹99 replaced a 60-piece pack at ₹499 that had underperformed. Quick commerce is now 40% of its domestic revenue [5].

A share reported for every channel. Snitch, the men's fashion brand, sells through its own site, marketplaces, 115 stores and a quick-delivery service. By its founder's unaudited figures, FY26 operating revenue was ₹900 crore, up about 80%. Online is about 60% of revenue and offline about 40%, and quick commerce is about 10% of online revenue. The article does not mention AI [6].

The pattern: they shaped the product or the pack for the channel, and they knew each channel's share. Neither needs software. Both need an owner who looks.

What the marketplaces now give sellers

Every large marketplace is building AI tools for its sellers. Most of what is known about them comes from the platforms themselves, so read the figures as marketing claims.

• Amazon says its seller base is over 20 lakh, that over 2 lakh sellers use its AI-powered listing tools, and that Seller Assistant interactions grew 7.5x year on year [7]. Inc42 reports that sellers using the assistant cut listing errors by 10% and routine operational work by nearly 70%, and that it answers in English or Hinglish [8].
• Meesho runs an AI voice system for sellers that handles up to 300,000 calls a day [8].
• Flipkart offers native tools for demand forecasting, pricing intelligence and trend analysis [8].

Sellers are using them. Snapdeal's survey of sellers on its own platform (sample size not stated) reports that 56% already use AI-powered solutions, and that 43% of those name product listings and content creation as their primary use [9].

Two cautions. Inc42 notes that no major marketplace publishes audited India data on what sellers earn back from advertising [8]. And the tool belongs to the landlord. It will tend to suggest more advertising, and a pricing tool will suggest a price. Use these tools for what they do well: they know the platform's own fields. Treat every suggestion about price, discount or ad spend as a suggestion. That decision stays with you (Chapter 18).

ONDC and UPI, briefly

ONDC. The network crossed 500 million cumulative transactions in July 2026, with 218 million in FY2025-26, close to nine million a month in June 2026 and more than 200,000 active retail merchants. A large part of that activity is mobility and public transport ticketing, at more than 370,000 public transport journeys a day [10]. The research for this book found no dated, named seller success story from 2025 or 2026. Worth a listing if your category is live on it. Not yet a channel to build a year's plan on.

UPI. In October 2026 UPI carried 25.47 billion transactions worth ₹31.53 lakh crore [11]. For a seller the news is about cost. NPCI had set a Merchant Discount Rate on select UPI merchant payments to begin on 15 October 2026, then postponed it, with a new date to be announced [11][12]. The proposed framework is 0.4%, capped at ₹300 per transaction, for large merchants on payments above ₹2,000, with a flat ₹5 in railways, telecom, insurance and fuel. NPCI estimates about 4% of merchant transactions would be affected [11]. Who counts as a "large merchant" was not confirmed. Ask your bank and your CA before changing anything.

Listings from your facts sheet

AI is good at structure and wording. It is dangerous with facts. So the system has four parts: facts from you, structure from AI, rules from the marketplace, and a check by a person.

The facts come from your product facts sheet, one row per product, containing only what you can stand behind. Chapter 14 shows how to build it. If you do not have one yet, write the row for one product before you use the prompt below.

1. Open the current style guide for your category in your seller account. Copy the title format, length limits, bullet rules, restricted words and image rules. These change and differ by category. An AI assistant does not reliably know the current version, so you must paste it.
2. Copy the product's row from the facts sheet.
3. Collect the words customers use: search bar suggestions, your own search term reports, the questions buyers ask.
4. Run the prompt.
5. Check the list of facts against the physical product. Then a person uploads it.

You are helping me write a marketplace listing. You arrange words. You do not supply facts.

Marketplace and category: [e.g. Amazon, Kitchen storage]
Category rules, copied from the current style guide: [paste title format, character limits, bullet rules, restricted words]
Product facts, from our product facts sheet (the only source of truth): [paste the product's row]
Words customers use when they search or ask: [paste list]

Write:
1. A title that follows the category format and length limit exactly.
2. Five bullet points. Each starts with a short benefit, then gives the fact that supports it.
3. A product description in short paragraphs.
4. Backend search terms, within the limit, with no words repeated from the title.

Rules:
- Use only the product facts I pasted. If a common specification is missing, write <<MISSING: name of the fact>> and do not guess.
- No price, offer, discount, delivery promise or warranty term unless it is in the facts.
- No "best", "No. 1", "guaranteed" or any claim the pasted rules restrict.
- No competitor brand names anywhere.
After the listing, list every fact you used, one per line, so I can check each against the product.

Check: every line of the checklist against the product in your hand, and the title length against the rule. Fill a <<MISSING>> from a supplier document or your own measurement, never from a competitor's listing.

For listings already live, run an audit before you rewrite anything.

Here are three things.
Marketplace rules for my category, copied from the current style guide: [paste]
Our current listing: [paste title, bullets and description]
The product's row from our facts sheet: [paste]

Give me three short lists:
1. Anything in the listing that may break the rules I pasted.
2. Any claim in the listing that the facts sheet does not support.
3. Important facts in the sheet that the listing leaves out.
Then suggest an improved title and bullets within the rules.
Use only what I pasted. Do not add facts. If you are unsure whether something breaks a rule, say so.

A margin sheet for every channel

Moms Home's lesson is to watch the margin as closely as the revenue. That needs one small sheet: a column for each channel, and a row for each thing that comes out of the selling price before the money is yours.

Example (fictional): A namkeen maker in Indore sells a 400 g pack on its own website and on a quick commerce app. The owner fills this in from last month's settlement report. All numbers are invented to show the layout, and leave out GST.

Per order	Own website	Quick commerce
Customer pays	₹200	₹200
Commission and platform fees	₹6	₹50
Advertising	₹20	₹30
Delivery and packing	₹45	₹10
Cost of goods	₹90	₹90
Left for you	₹39	₹20

The owner now knows what each order leaves behind on each channel, and can decide whether a smaller pack, a kit or a lower ad budget would change the last row.

Your rows will differ by platform: returns, storage, funded discounts and penalties belong in the sheet if you pay them. Fill it from settlement reports, not from memory. An AI assistant can set up the sheet and write the formulas. Do not ask it to do the sums itself, or to tell you your price. Chapter 18 explains why.

The funnel: find the biggest drop

Every online sale passes through the same stages. The product is shown. The page is opened. It goes into a cart. It is ordered. Sometimes it comes back. Buyers leave at every stage. Baymard Institute puts the average documented cart abandonment rate at 70.22%, an average of 50 studies [13]. Your own figure will differ. The useful comparison is between your own products, and between this month and last.

Example (fictional): A kitchenware brand in Moradabad compares two tiffins on one marketplace for one month. All numbers are illustrative.

Stage	3-tier tiffin	2-tier tiffin
Shown in search	40,000	38,000
Page opened	1,200	1,140
Added to cart	96	34
Ordered	60	21
Returned	4	6

Both are opened by 3% of the people who see them. But 8% of visitors add the 3-tier to the cart and only about 3% add the 2-tier. The 2-tier also comes back far more often. The owner opens the page and finds no photo that shows its size. Buyers either cannot judge it and leave, or guess wrong and return it.

Each drop points to a different part of the listing.

Below are the funnel numbers for [number] of our products on [marketplace] for [period], copied from our seller reports. There is no customer personal data in them.
[paste a table: product, times shown, page views, added to cart, orders, returns]

1. For each product, work out the percentage that moves from each stage to the next. Show the division for every figure so I can check it.
2. For each product, name the stage with the weakest percentage compared with our other products.
3. For each weak stage, list what usually affects that stage on a listing (for example main image, title, bullets, delivery charge) as questions for me to check. Do not tell me the cause. You cannot see the listing.
4. Do not recommend any price, discount or advertising budget.
Use only the numbers I pasted. If a number is missing, write "not given".

Check: two of the percentages on a calculator. Then open the listing on a phone, as a buyer would, and answer the questions yourself. Fix one thing, wait a full period and compare again.

From my desk: NEWDRU, my consumer products brand, is sold on Amazon, Flipkart, Blinkit and its own website. So this chapter is written from the seller's side of the screen. My advice to any owner in that seat: know what each channel leaves you per order before you chase its growth.

Mistakes to avoid

• Listing your existing pack on quick commerce without asking whether the channel wants a different size, price point or kit.
• Reporting channel revenue and never the contribution per order.
• Letting AI write a listing from "what such products usually have" and not from your facts sheet.
• Asking an AI assistant for the marketplace's rules. Paste the current style guide yourself.
• Accepting a platform tool's suggested price or ad budget without doing your own arithmetic.
• Changing five things on a listing at once, so you never learn which one worked.

Do this week

1. Build the margin sheet for your top product on every channel you sell it on, from last month's settlement reports. Circle the lowest "left for you".
2. Run the audit prompt on your best-selling listing with the current category rules pasted in. Fix every unsupported claim it finds.
3. Pull the funnel numbers for your top five products, run the funnel prompt and pick one weak stage on one product to fix.

Sources

1. Inc42, feature on how Moms Home built its quick commerce business (founder-reported figures), 28 October 2026. https://inc42.com/startups/how-moms-home-built-a-₹10-cr-quick-commerce-business-by-rethinking-baby-essentials/
2. Kotak Neo, report on Eternal's Q2 FY27 results (Blinkit figures), 21 October 2026. https://www.kotakneo.com/news/stocks/eternal-q2-results-profit-rises-revenue-growth-blinkit-adds-stores/
3. Kotak Neo, report on Swiggy's Q2 FY27 results (Instamart figures), 23 October 2026. https://www.kotakneo.com/news/stocks/swiggy-q2-results-loss-narrows-revenue-rises-instanmart-loss-widens/
4. Outlook Business, coverage of Zepto's IPO filing (FY26 figures), 10 June 2026. https://www.outlookbusiness.com/markets/can-zepto-catch-blinkit-what-its-ipo-filing-reveals-about-indias-quick-commerce-war
5. Inc42, feature on Pee Safe's quick commerce business, 21 October 2026. https://inc42.com/features/how-pee-safe-built-a-profitable-quick-commerce-business-that-drives-40-domestic-sales/
6. Inc42, report on Snitch's FY26 figures (unaudited, from the founder), 10 April 2026. https://inc42.com/?p=554737
7. Amazon India, press release on its seller base and seller tools (vendor-published), 9 September 2026. https://press.aboutamazon.com/in/2026/9/amazon-indias-seller-base-grows-to-20-lakh-marking-strongest-growth-in-three-years
8. Inc42, feature on Amazon, Flipkart and Meesho seller-side AI, 18 August 2026. https://inc42.com/features/amazon-flipkart-meeshos-ai-focus-shifts-to-seller-side-stacks/
9. MediaNews4U, report on the Snapdeal Bharat Seller Report 2026 (survey of Snapdeal's own sellers; sample size not stated), 1 June 2026. https://www.medianews4u.com/56-of-online-sellers-now-use-ai-tools-to-grow-their-business-snapdeal-bharat-seller-report-2026/
10. Cross-Border Magazine, report on ONDC reaching 500 million transactions (figure attributed to DPIIT), 7 August 2026. https://cross-border-magazine.com/ondc-500-million-transactions-india/
11. IANS via Free Press Journal, report on October 2026 UPI transactions (NPCI data), 2 November 2026. https://www.freepressjournal.in/business/upi-transactions-hit-2547-billion-in-october-value-soars-to-3153-lakh-crore
12. Free Press Journal, report on September 2026 UPI transactions and the planned Merchant Discount Rate, 1 October 2026. https://www.freepressjournal.in/business/upi-transactions-fall-18-in-september-to-2407-billion-ahead-of-october-15-mdr-rollout
13. Baymard Institute, "70.22% – average documented online shopping cart abandonment rate" (an average of 50 studies), checked November 2026. https://baymard.com/lists/cart-abandonment-rate