By Varun Mahajan, FounderPublished Last updated Part of the quick commerce data and api guide
What the API returns today
These are the commerce datasets most integrations start with. Each row is the catalogue card as /data shows it: its status, what it covers and when it was last updated.
| Dataset | Status | Coverage | Freshness |
|---|---|---|---|
| Prices and stock by pin | Live | 6 sources · 677 pins held · fresh read at any pin on Blinkit · 1 behindBehind: BigBasket. See each source. | Daily, last updated 31 Oct 2026, 1:39 pm IST |
| Prices and listings | Live | 8 sources · 216 pins held and national · fresh read at any pin on Ajio | Daily, last updated 31 Oct 2026, 1:41 pm IST |
| Delivery time by pin | Live | 4,472 pin-days of delivery times: Blinkit daily at 35 pin codes since 20 Sep 2026, and six apps at up to 1,247 pin codes from partner reads and sweeps (16 Sep to 16 Oct 2026) | Daily (Blinkit searches); the other sources are historical, last updated 22 Oct 2026, 8:07 pm IST |
| Dark-store locations and catchments | Live | 13,736 dark stores on 7 apps, in 688 cities, serving 5,482 pin codes (as of 31 Oct 2026) | Weekly sweeps; newest seen 29 Oct 2026, last updated 29 Oct 2026, 10:45 am IST |
| Price per unit | Live | 13 retailers · checked on 3,584,437 readings (8 to 30 Sep 2026)Built, and checked reading by reading against the previous system (3,584,437 readings for 8 to 30 Sep 2026, no mismatch). The daily run is not switched on yet. | Not collected yet |
| Stock state | Live | 271 pins held and national | Daily, last updated 30 Oct 2026, 5:15 am IST |
A Live card with a note that its daily run is not switched on yet has been built and checked, but is not refreshed every day. Read the note before relying on it for a current answer.
Catalogue discovery is distinct from retrieving the records. A card tells you what an entry contains, where it is covered and how recently it was updated; it is not a sample of everything Rewalto holds. Check whether records are kept over time or read fresh for a request, and review the source and usage information before planning redistribution or a customer-facing feature.
Which sources it reads
- Quick-commerce prices and stock by pin: Blinkit, Zepto, Instamart, JioMart, DMart and BigBasket.
- Marketplace prices and listings: Amazon, Flipkart, Myntra, Ajio, Nykaa, Purplle, Tata 1mg and Tata CLiQ.
Rewalto reads public pages itself; it is not any app's or store's own API, and none of them supplies or endorses this data. Each reading keeps the source and time it was read. For each app on its own, see the commerce datasets; for where each app delivers from, see the Dark-store map.
What is live, in preview and planned
Live means a request can run today. Preview means it exists but some parts are not ready for production use. Planned means it is described in the catalogue and cannot be requested yet. Labels change as datasets move, so this section reads them from the catalogue rather than repeating a list written on one day. A planned dataset is a description of intent, not an offer you can buy.
- Live (21): Prices and stock by pin, Prices and listings, Ratings over time, Price per unit, One product across retailers, Delivery time by pin, Launches, Brands and aliases, Search and category ranks, Stock state, Search rank and sponsored slots, Unit velocity, Dark-store locations and catchments, Badges, Share of search, Sponsored share, Price parity, Competitor ad launches, Brand share of shelf, Out-of-stock rate and Festive discount tracker. Access through AI agents (MCP, preview) is in preview.
- Preview (2): Brand to company and Promo emails.
- Planned (9, not yet available to request): Where to buy by pin, Sellers and buy box, Review text, Assortment per dark store, Offers and bank offers, Delivery promise by pin, Co-purchase, Demand and Delivery promise.
These are the commerce datasets as the catalogue labelled them when it was checked on 31 Oct 2026, 1:43 pm IST. Across every family, it listed 30 live, 3 in preview and 24 planned.
Make a first call
Rewalto's REST address is https://api.rewalto.com. Start without a key: the version route needs no credentials, and the public catalogue is free, so you can confirm that a dataset is live before you spend anything.
curl --fail-with-body "https://api.rewalto.com/v1/version"
curl --fail-with-body "https://api.rewalto.com/v1/catalogue/products?status=live"Check the version against the integration note before relying on a capability. For price and stock at a pin code, see the worked example in is there an API for India's quick-commerce apps. The API quickstart covers authentication and the data reference lists the documented routes.
Keep private keys in server configuration, never in browser code, a public prompt or a shared report. The site, CLI and MCP use the same core API, credentials and approval rules, so an agent connection does not bypass scopes or credit review. To connect an agent through MCP (preview), read the API, CLI and MCP reference first: client setup stays unavailable until its integration note is published.
What a read costs
Credits are for the API, MCP (preview) and CLI; the dashboard never spends them. Each dataset publishes its rate in credits per unit on /data, and you pay only for data delivered. A request that returns nothing costs nothing. Every account gets 1,000 free credits a month. Free credits reset on the first of each month; paid credits last twelve months.
Plan pagination before a wide extraction: a small first page validates your integration but does not tell you the cost of the whole set. When no published rate applies, review the quote and approve it before anything runs. Keep the receipt and request reference with the delivered records, without copying private content into logs. Decide who may expand a query, follow later pages or approve work beyond the first scope: a record returned while debugging is still delivered data. The pricing page explains how credits work.
Write a field checklist your application can enforce
Separate required fields from useful extras. For a commerce comparison this might be product identity, pack size, retailer, location, listed price, availability and collection time. Use the published field reference rather than assuming a familiar column name exists.
Define missing values before building calculations. An absent price is unknown, not zero. A missing availability check does not prove that a product is unavailable. Keep identifiers as exact strings, including leading zeroes, and keep original values beside your own display labels so a reader can trace a comparison back to the supplied records.
Treat location and time as part of the record
India-wide coverage is not coverage of the pin codes your question concerns. Quick-commerce results differ by pin code; marketplace listings have a different location basis. Confirm how each dataset represents geography before joining records from different sources.
Collection time, catalogue update time and the time your application received a response are different events. Keep them separate, and document the time zone when comparing periods. A gap in a series should stay visible rather than being filled with a price from another place or day.
When you hand a comparison to another team, include the selection rules: which pin codes or marketplaces were requested, which returned records and which remain unknown. Put the limitation beside the comparison, where the reader can see it, rather than only in an implementation note.
Make errors and incomplete results useful
Show authentication failures, missing permissions and unavailable data explicitly. Do not replace a failed call with a successful-looking empty table. After an uncertain write, check the known resource before trying again; a timeout does not establish that nothing happened.
Before shipping, test missing fields, empty results, later pages, expired credentials and an unavailable service. Review the error guide and the coverage and freshness checks together, then accept the specific journey your users need.
What it cannot do
- It does not report sales or demand. Listings and prices are not units sold; Unit velocity on /data is a lower bound from stock counters, not a sales figure.
- It does not place orders or act on a seller account. The documented data routes only read.
- It does not guess. A missing reading is unknown, and an unread pin code returns no record.
- It does not make marketplace geography match quick-commerce pin codes. Check each dataset's location basis before joining them.
Beyond commerce
The same catalogue also lists companies, startups, investors, public records and brand signals, through the same API, credentials and credit rules. See the Rewalto data catalogue for every family, and marketplace product data or shelf data by CSV or API for how to read the commerce records well.
Questions people ask
Is there an ecommerce data API for India?
Yes, for reading data. Rewalto's API returns price, MRP, discount, stock and listing records from Indian quick-commerce apps and marketplaces, each with its pin code where the source has one and the time it was read.
Does it work by pin code?
Yes for quick commerce, where prices and stock differ by pin code. Marketplace listings carry their own location basis, so check the field before joining the two.
How fresh is the data?
Each dataset shows its own freshness on /data. When the catalogue was checked on 31 Oct 2026, 1:43 pm IST, quick-commerce prices and stock were daily, last updated 31 Oct 2026, 1:39 pm IST, and marketplace prices and listings were daily, last updated 31 Oct 2026, 1:41 pm IST.
How much does it cost?
Reads are charged in credits for data delivered. Prices and stock cost 1 credit per 10 data points, and dark-store locations cost 5 credits per dark-store record, as the catalogue showed on 31 Oct 2026, 1:43 pm IST. Every account gets 1,000 free credits a month, and a request that returns nothing costs nothing. The pricing page explains how credits work.
Is this the apps' or marketplaces' own API?
No. Rewalto reads what each app or store shows shoppers on its public pages. It is not a partner of any of them and is not their own feed.
Can an AI agent use it?
Yes, through the same API, a CLI, or the MCP server (preview) at mcp.rewalto.com/mcp. The same credentials, permissions and credit rules apply to all three.
What happens when there is no data?
You get an empty result or an unknown state with a reason, never a guess, and it costs no credits.
Start with the question you need to answer
Write a short acceptance statement before choosing a dataset: which decision the data will support, who will read the result, the minimum pin codes or marketplaces and the period. That gives developers and reviewers one definition of success and makes an honest partial result easy to recognise.
Different decisions need different records. A purchasing comparison needs equivalent products and prices read under comparable conditions. A screen showing recent availability needs the pin code and reading time beside each result. A historical analysis needs a documented period and an explanation of gaps. These are different requirements even when they concern the same product, so name the one you have before choosing a dataset.
If the data you need is missing, ask Finder with the fields and scope. To check one brand at one pin code without code, use the free pin-code availability check.