India store openings tracker

How this tracker is built

Each row is read from the headline and summary of one live, source-cited RetailIntel signal by a fixed rule set, with no model call. A field the text does not state stays empty. Brand, city, format and counts are spans of that text, or lookups keyed on one (city → state, a well-known locality such as Powai → Mumbai, “flagship” → flagship store).

In scope: physical, consumer-facing locations in India — stores, flagships, boutiques, showrooms, experience centres, dark stores, kiosks, cafés and restaurants, cinemas, dealerships, salons, pharmacies. Out of scope: hotels, malls, offices, plants, warehouses, bank branches, clinics, online stores, fuel pumps, distribution reach (“adds 50,000 outlets”) and openings abroad.

An opening is a report that a location opened; an announced plan is a brand saying it will open stores, with the target it stated. A brand’s parent and the brand reporting the same stores in the same month are one row, and so are several reports of one opening: a parent and its store banner reporting one opening in the same city, on the same day, at the same network size count once, and a company’s legal name and its trading name (Avenue Supermarts, DMart) count as one brand. A store count is the number the headline gives for this opening, never the brand’s whole network.

A period total is a report of stores added over more than one month (“adds 18 stores in H1 FY27”, “Q2 FY27 … adds 23 stores”, “20 stores this year”, “26 stores in 60 days”). It sums openings rather than reporting one, so the CSV lists it with its period (status period_total) and it is never counted as a month’s, a city’s, a brand’s or a format’s openings. The line is one month: “five stores in 45 days” is a period total, “three stores this week” is an opening. A count the report splits into stores already open and stores still to come (“12 stores, three already operational and nine opening in the coming weeks”) is an opening of three and an announced plan of nine. A venue the report places the opening in (“a new store at DLF Summit”) is never read as the brand, and an outlet of a restaurant or café brand is filed as a restaurant or café. These rules (October 2026) moved 23 rows, 2,825 stores, out of the monthly opening counts; October’s openings on record went from 38 to 33.

Month = the month the text states, else the month of the source report. When our pipeline could not date a report, the date comes only from the cited article itself (its URL or its own publication date), never from when we collected it. Archive articles re-published years later therefore fall outside the window.

Accuracy is measured by hand on fresh samples the rules were never tuned on: 40 of 40 openings correct (three of them name the parent company rather than the store banner, e.g. Arvind Fashions for a Megamart store) and 17 of 20 announced plans (October 2026); 22 of 22 openings reported in the week to 13 October 2026, three of them with one wrong field (a missing city, a count that included stores not yet open, a restaurant filed as a store; all three are fixed). After the period-total rules, the 14 openings no earlier sample had read were all correct, two with an empty store count (a single store whose size comes before the word, “21,000 sq ft … store”). Found an error? Report it on the corrections page.