Answers computed from the dataset itself, snapshot 2026-08. Where an answer quotes a number, that number is read from the same file the rest of the site renders from, so it cannot drift. Where we do not know something, the answer says we do not know it.
Fourteen questions, including the ones with unflattering answers.
Our snapshot of the government register holds 4,53,192 companies, taken in 2026-08. The government's own facet counts, computed over the whole register, return 4,67,663. Ours counts companies after duplicate profiles are merged, theirs counts profiles, and we also drop records we may not republish. The state-by-state reconciliation prints every gap, including the ones that do not flatter us.
2,30,545 of our records, about 51%, carry DPIIT recognition. Recognition is an eligibility status granted on application: it opens the door to income tax benefits, self-certification under certain labour and environment laws, and faster IPR support. It is not an audit of the business, not proof that the company trades, and not a financial assessment. We show the status and the DIPP number as the portal publishes them, and never assert either ourselves. How we handle it.
We cannot tell you, and no page here implies otherwise. A registration is a form somebody filled in. We hold no liveness signal of any kind: no filings feed, no revenue, no headcount, not even a check that the website on file still resolves. A large share of these entities are dormant, never traded or already wound up. If a page here says a company exists, it means a record exists. The methodology says so at length.
Delhi, with 43,654 records, ahead of Bengaluru at 34,244. Read those as districts rather than city limits: Startup India records a district, so we group districts into the metro they belong to only where that is unambiguous. 698 districts clear 25 companies and get their own page. The full ranking.
Because a district is usually all the source gives us. 87% of our coordinates are district centroids, 50,073 records resolve to a locality, and only 4,345 have a street-level coordinate. Those are the only records ever drawn as a pin. Separately, we never publish the address line itself: registered office addresses in company-registry data are frequently founders' homes.
No. Indian Startup Map is an independent publication by Apexlayer Technologies. We are not DPIIT, not Startup India, and not affiliated with the Government of India. We take open data the government already publishes and make it readable, with attribution and with the provenance shown on the record. Nothing here is a certificate, an endorsement or a verification, and nothing here should be used for due diligence.
Less than half of it. 2,28,415 of 4,53,192 records carry something beyond a name, a district, an industry and a stage: a description, a website, a CIN, a logo. The remaining 50% are registry rows and nothing more. We publish them, mark them so search engines do not index them, and tell you the proportion rather than padding them out with generated prose.
The government's own taxonomy sorts them into 56 industries. The largest is IT Services with 48,076 records, then Healthcare & Lifesciences with 35,688. Industry, sector and stage are close to complete across the register, which makes them the most reliable fields we hold. All three are declared by the applicant, not assessed by anyone. Browse by industry.
The Startup India Seed Fund Scheme has approved Rs 755.3 crore across 175 incubators in 26 states. Rs 526.4 crore of that, 70%, has been disbursed. Both figures are published by the scheme, and almost nobody puts them next to each other. The ledger shows approved against disbursed for every incubator, largest first.
Indian government open data: the Startup India (DPIIT) register, the Startup India Seed Fund Scheme, iStart Rajasthan and Kerala Startup Mission. Records are normalised into one schema, deduplicated by CIN, then website hostname, then normalised name plus city, geocoded, and republished with attribution. Every source with its licence, including the proprietary ones we use only to match records and never show.
Yes. The counts, aggregates, charts and rankings on this site are free to reuse, including commercially, with attribution. We ask two things. Link to the page you took the figure from rather than to the home page, so a reader can check it. And quote the snapshot, because this is a point-in-time sweep. The licence, and a citation you can paste.
We re-sweep the register roughly monthly and publish each sweep as a dated snapshot. This one is 2026-08, generated 2026-08-19. Nothing on the site updates continuously, and no figure here should be quoted without its snapshot: a number without a date silently becomes wrong. Corrections made at the source portal reach us on the next sweep. How the sweep works.
Everything here derives from public government records, so the authoritative fix is at the source portal, and we pick it up on the next snapshot. To flag something on our side, write to trust@indianstartupmap.com. Where a source is wrong we annotate rather than silently overwrite: the page will say what the portal says and what we believe. Trademark owners can have a logo removed on request.
No, nowhere on the site. Director names from the corporate registry mirror are held internally for matching records and never published in any form. Those lists routinely include spouses and parents added to satisfy the two-director requirement, who never worked at the company. Registered office address lines are treated the same way. What we do and do not collect.
The methodology is the long version: sources, precision tiers, the two different dates on a record, and an explicit list of what we do not know. For anything else, or to tell us a page is wrong, write to trust@indianstartupmap.com.