Dark Store Index · Open data

The shelf data behind the reports, free to download and reuse.

430 brand rows and 54 search rows, read from the public shelf of individual dark stores on Blinkit, Swiggy Instamart, Zepto in Bengaluru and Delhi NCR. Licensed CC BY 4.0: reuse it with credit.

Read 23 July 2026 – 1 August 2026 / Published 20 November 2026 / CC BY 4.0 /

Download

Current version (read dates 23 July 2026 – 1 August 2026)CC BY 4.0
FileWhat it holdsRowsSize
dsi-brand-shelf-asof-2026-08-01.csvOne row per brand per search per city and platform430230 KB
dsi-search-summary-asof-2026-08-01.csvOne row per search per city and platform: stores read, brands seen, results depth, price range, ad coverage5429 KB
dsi-shelf-data-asof-2026-08-01.jsonBoth tables plus the method note, in one JSON file484159 KB

Plain-text summary: README.txt. File names carry the newest read date, and earlier versions stay online at the same address.

Licence and attribution

This data is licensed under CC BY 4.0. You may copy, redistribute, adapt and use it commercially, as long as you credit it. Use this credit line, with the date of the read:

Source: The Dark Store Index, [date of the read]. Link to darkstoreindex.com where you can.

The date matters more than the name: these shelves change daily. Every row carries its own read date, so quote the date on the row you used. Citation formats are on the how to cite page.

What was read, and when

Read dates by city and platformMeasured
CityPlatformRead dateDark stores
BengaluruBlinkit1 August 20269
BengaluruSwiggy Instamart23 July 20269
BengaluruZepto1 August 20269
Delhi NCRBlinkit24 July 20269
Delhi NCRSwiggy Instamart24 July 20269
Delhi NCRZepto24 July 20269

Different city and platform sets were read on different days, so this is not a single-day snapshot. Searches covered coffee, protein & energy bars, baby care, three searches per category. One city on one day is a snapshot, not a trend.

Columns

Field dictionary
ColumnMeaning
read_dateDate the dark stores were read (YYYY-MM-DD). Cite this date.
city, platformBengaluru or Delhi NCR; Blinkit, Zepto or Swiggy Instamart.
category, keywordThe search a shopper would type, and the category it belongs to.
brandBrand as it appeared on the shelf (brand table only).
stores_present / stores_read / coverage_pctDark stores where the brand appeared, dark stores successfully read, and the percentage.
best_rankBest position the brand reached in any store read.
lowest_rank_seenLowest position observed within capped results. A floor, not a catalogue rank.
sponsored_slots_ownedPaid placements the brand held across the stores read. Not rupees spent.
distinct_brands, results_depth_*, stores_with_any_ad, stores_with_no_ad, price_*_inrSearch-level figures (search summary table).
method_note, license, attributionRepeated on every row so the terms travel with the data.

Method and limits

  • Public shelf only. Counted from what a shopper sees. No seller accounts, no brand-supplied data, nothing modelled.
  • Per dark store, never a national average. The brand table lists only the top 8 brands per search by store coverage, so a brand missing from a search may still have appeared lower down.
  • Ranks are a floor. Platforms cap how many results a store returns, so lowest_rank_seen limits what we could read. It is not a catalogue rank.
  • Ads are slots owned, not spend. Bids are not public, and we never convert slots to rupees.
  • A store we could not read is excluded, never counted as absent.

The full method, and what the Index deliberately does not claim, is on the About page. Spotted a wrong number? Write to hello@darkstoreindex.com and it will be corrected in place with a note.

Citing a number?

Suggested formats for APA, MLA, Chicago, Harvard and BibTeX, plus a one-line credit for charts and posts.

How to cite →