For investors, analysts and researchers

Research-grade intelligence on India’s quick commerce.

For people who study the channel, not one brand. The methodology states how each figure is counted, so it can be checked, and quoted.

Download the sample file (Excel): four sheets, a few rows of each table, laid out like a file the product exports.

1,019brands were on the shelf in skin care.

Out of
the dark store shelves of Blinkit, Zepto, Swiggy Instamart, Flipkart Minutes and BigBasket, store by store
Rule
A brand is counted once in a category, on however many apps it was listed.
Source lineSource: Shelf (seetheshelf.com). Brands on the shelf in skin care on Blinkit, Zepto, Swiggy Instamart, Flipkart Minutes and BigBasket, the dark store shelves of Blinkit, Zepto, Swiggy Instamart, Flipkart Minutes and BigBasket, store by store. A count of what the shelves listed, not of sales.

From the read

Six findings, each with what it is out of.

These are figures about the channel, not about any one brand or any one app. Under each is what it was counted over, and the rule it was counted by.

Paid slots in search

4% to 36% of the result slots were marked as paid, depending on the app.

Blinkit 15%
Zepto 36%
Swiggy Instamart 11%
BigBasket 4%
Flipkart Minutes 16%
JioMart 19%
Out of
Every result slot on each app’s searches
Rule
A result the app marked as paid for. We count where it appeared; we never see what it cost. Method

A brand’s own name

Another brand held a paid slot on 51% to 98% of brand names, depending on the app.

Blinkit 81% of brand names
Zepto 98% of brand names
Swiggy Instamart 78% of brand names
BigBasket 51% of brand names
Flipkart Minutes 57% of brand names
JioMart 80% of brand names
Out of
Searches for a brand’s own name, on each app
Rule
Another brand held a paid slot on a search for a brand’s name, in five or more cities. It says where the slot appeared, not what that brand chose. Method

How crowded a category is

Skin care had 1,019 brands on the shelf.

Skin care 1,019 brands
Medicines 818 brands
Supplements and protein 584 brands
Hair care 568 brands
Home appliances 539 brands
Party and festive 475 brands
Out of
the dark store shelves of Blinkit, Zepto, Swiggy Instamart, Flipkart Minutes and BigBasket, store by store
Rule
A brand is counted once in a category, on however many apps it was listed. It is a count of what the shelves listed, not of what sold.

The same pack on two apps

Of the products matched on two apps, 43% were lower on one, 30% on the other and 27% were level.

Lower on one app 43%
Lower on the other 30%
Level 27%
Out of
The products matched as the same product and pack on both apps
Rule
Two apps are compared only on products matched as the same product and pack on both, by median price over the stores on each. A product counts as lower on one app when it is more than 1% lower. A count, not a saving. Method

What a store network looks like

On one app, 63% of the dark stores state hours that span the whole day.

Open 24 hours 63% of its dark stores
Pharmacy stock reaches it 39% of its dark stores
Out of
The dark stores of one app
Promise
The median delivery promise was 8 minutes
Rule
The hours the app stated for the store span the whole day. A store with shorter stated hours closes at night; a store with no hours read is in neither group. Method

Stores, app by app

11,234 stores in 984 cities, each one read on its own.

Blinkit 2,650 dark stores
Zepto 1,406 dark stores
Swiggy Instamart 1,318 dark stores
Flipkart Minutes 1,199 dark stores
BigBasket 881 dark stores
Amazon Now 863 stores
JioMart 2,917 stores
Counted
Each app’s stores, each read on its own at its own delivery area. JioMart’s are Reliance’s own stores, and Amazon Now’s include partner stores
Rule
A store is a store id as the app issues it, counted once, on its own app.

In the product

What you can study.

A research account opens on the market, by category. It follows any brand it wants to look at from outside.

Sizing a category · Looking at a brand from outside · Quoting a figure

A category, on every app
How many products and brands, how much the top three hold, who leads, the median price per 100 g and how much the app’s own labels take. One app at a time, never added across apps.
Any brand, from outside
Follow a brand and read it as its own team would: the stores that list it, by state and city, its prices beside the brands it competes with, its place in search, and the brands that held a paid slot on its name.
Paid slots, by app and by keyword
Which brands held the paid slots, on which keywords and in how many cities, and how much of the first ten results was paid.
Each app’s dark stores
Counts by city and state, the hours a store states, the minutes it promises, and how deep a category runs in a typical store.
Prices, across apps and between reads
The same pack on every app that lists it, and how prices moved between two reads, for the products seen in both.

Method

Built to be checked, and quoted.

A figure is only as good as what a reader can check about it. These are the rules every figure in the product is printed under.

  1. 1

    How each figure is counted

    The methodology says it, rule by rule.

  2. 2

    Files keep the day

    A file carries the day it was read, so files of different reads stack.

  3. 3

    A definition for every term

    206 terms are defined in one sentence each, and the same sentence appears beside the figure on screen and on the first sheet of a file.

  4. 4

    A blank is not a zero

    A figure we do not have, or would not print, is a hatched cell (drawn with stripes) with its reason. A brand is named on a keyword only when it held a paid slot in five or more cities, and brand figures are withheld where too many listings came with no brand.

  5. 5

    A file that explains itself

    An exported file opens on a sheet that states the read and the rule. The sample file on this page is laid out the same way.

  6. 6

    Corrections in public

    A change to how a number is counted is listed in the changelog when it takes effect, and so is every correction a reader asked for.

How we read the apps · The glossary · The changelog · Status and coverage

Plans

What each plan includes

A research account has no brand of its own. The brands you follow count the way rivals do on a brand account, so the same plan limits apply.

What a research account seesFreeno cardPro₹29,999 a month
Rivals watched at once1Every brand in your categories
A rival’s search and ads: its keywords, its paid slots and its citiesIts totalsEvery rival, in full
Cities, each counted city by city1 of your choiceEvery city
Categories and the apps side by side: share of shelf, who leads, own labelsCounts, the first 5 brandsEvery brand ranked
Excel and CSV of any tableNot on this planUp to 300,000 rows a month

Prices in rupees, per month, GST included. Pay once for 30 days, or turn on autopay and stop it any time. Compare every plan

One reader: Pro Research.

For investors and researchers

Pro Research

₹29,999 a month

GST included

The whole market, and who is winning.

Refreshed every day, and every hour in your target cities once set up

  • The same as Pro, with no brand of your own:
  • Where it is heading: the same products’ prices, read to read
  • Who wins search, and who held paid slots on whose name
  • What shoppers rate most, where the app prints it
  • Which app lists the most, and where each category runs deepest

Or start free.

Say what you study and the market opens by category, with each category’s leading brands. Move to Pro Research when you need the research screens, every brand ranked, every city, the whole record and the files.

An institution.

A fund, a research desk, a university or a newsroom usually needs what a card cannot buy: more than one reader, a written licence to republish, files on a schedule, an invoice to a purchase order. Tell us what you study and we agree it in writing.

Talk to us

Every store of the 22 apps, with its position, its details and each week’s changes, is a product of its own beside the plans: Location intelligence. A Shelf plan does not include it.

Questions from people who study the market

Can I use Shelf without a brand of my own?

Yes. A research account has no brand on purpose. You say which categories you study, and you can follow brands. The brands you follow count the way rivals do on a brand account, so the same plan limits apply.

May I quote a figure in a note, a paper or an article?

Yes, with attribution to Shelf (seetheshelf.com). The product gives you a source line with how it was counted. Republishing the data in bulk, or reselling it, needs a written agreement.

Is this a panel, a survey or an estimate?

Neither a panel nor a survey. It is read from the apps’ own pages, store by store and search by search. A figure we estimate says so in its first words.

Can it tell me how much an app or a brand sells?

No. It shows what is on the shelf, in search and in the paid slots. Sales, revenue and spend cannot be read from a page, and Shelf prints none of them.

Is any of it investment advice?

No. Shelf publishes measurements. It makes no forecast, and it holds no view on any company or its shares.

Do you sell dark store counts and positions?

Yes, as a product of its own: Location intelligence, every store of the 22 apps refreshed every week, with each store’s position, address and details, and what changed each week. It is bought on its own or beside Pro Research.

Can I see the data before I pay?

Yes. The sample file on this page is a small Excel workbook laid out like a file the product exports: its first sheet states what was read and the rule for every term, and each table shows a few rows with some cells held back. A research account is also free to set up.

We are a fund, a research desk or a university. How do we buy?

Tell us what you study and who will read it. One reader can pay by card. For more than one reader, a written licence to republish, files on a schedule or an invoice to a purchase order, we agree it in writing.

Read the market.

A research account is free to set up, and takes about a minute.

No card, no seller login and no brand needed.