Methodology

How we build the catalog, and how to check our work

A grant directory is only worth the freshness of its worst entry. This page is the standing answer to a fair question: why trust these numbers? It covers where the data comes from, how often it is checked, how eligibility is scored, and the things we refuse to do. If you find something here that does not match what you see in the product, that is a bug and I want to hear about it.

Where the data comes from

Four sources, in descending order of how much we trust them.

1. The scheme's own portal. Every figure quoted in our guides and weekly posts is read off the programme's official site on a recorded date. Not from a press release, and not from another aggregator. This matters more than it sounds: search for NIDHI PRAYAS and you will be told repeatedly that it pays up to ₹10 lakh. The programme portal says up to ₹20 lakh at a PRAYAS Centre and up to ₹40 lakh at an Advance Centre. The lower figure appears to be carried over from an earlier version of the scheme and copied between articles ever since.
2. Curated lists and official listings. Programme sets kept by hand, deepest for India and the Gulf, plus a few listings read directly from the funder: the Startup India funding page, the EU Funding and Tenders portal and SBIR.gov.
3. A weekly AI discovery pass. Every week it covers India's central schemes, four separate passes across Indian states so smaller missions are not crowded out by the same six names, CSR funds, and three passes across the Gulf. The rest of the world rotates on a slower cycle. This pass works from a model's knowledge, not from a live page, so we treat what it finds as a candidate. A row must carry a web address to be stored, and that address is not trusted until the link sweep has checked it. A row that claims more money than the scheme's own portal publishes is dropped. A deadline that has already passed is not refused on arrival; the daily hygiene pass removes it, as below.
4. A browser extractor. Once a week it reads a list of official sites with a real browser and proposes new programmes. A person reviews them before anything is added.

Sources we stopped using, and why. The World Bank, NIH RePORTER and UKRI feeds returned real data, but it was lending to governments and research already funded, nothing a startup can apply to. grants.gov came out on 16 October 2026: its open calls are US federal research opportunities that need a US entity, and none is winnable by an Indian or Gulf founder. The rows those feeds left behind were deleted, and the deletions are recorded so they cannot come back.

The official portals stay the source of truth for every scheme rule. How Incubateer differs from Startup India and myScheme, and when to use each, is on the comparison page.

How a dead programme leaves

About half the "official" grant pages on the internet are gone, and government portals move or retire without redirects. So links are re-checked on a rolling weekly sweep, and the rules are deliberately conservative:

• A link that fails DNS, or returns 404 or 410, gets one strike. It is deleted only if it fails again on a later sweep.
• Timeouts, connection refusals, 403s and 5xx are never grounds for deletion. Those are usually geo-blocks, bot defences or a bad afternoon, and deleting a live programme is worse than carrying a stale one for a week.
• A grant with a real dated deadline is removed a week after that date passes. Rolling, annual and twice-yearly schemes carry no expiry and are left alone, so a programme like BIRAC BIG, which opens on 1 January and 1 July, survives the pass.
• Copies of the same programme, listed under slightly different names, are merged into one by the daily hygiene pass. On 16 October 2026 we also ran a larger one-time merge, reviewed and approved before it ran, that cleared copies which had built up over months. Anyone who had saved or applied to a copy keeps their record on the programme that survived.
• When a government portal moves, the programme is pointed at its new address and checked again, rather than deleted. A batch of moved portals was repointed this way on 16 October 2026.
• A programme we remove stays removed. Deletions are recorded, so a redeploy cannot quietly resurrect a dead link, and discovery will not re-add the same name for 90 days.

The consequence worth stating plainly: the catalog total moves down as well as up. In one recent sweep, 99 of 800 links checked were confirmed dead and deleted. A July purge removed 1,771 in one pass.

Those are past numbers, typed in when this page was written. Here is a current one, read from the catalog when this page loads, with a conservative floor baked in for anyone it cannot reach: 2,400 programmes removed since 22 September 2026, which is when we started keeping a record of removals rather than just making them. We say the date rather than "in the last 30 days" because the record is younger than that window, and a number presented against a window we were not watching for all of would read six times larger than the real rate. If the catalog cannot be reached, the figure above is the last published floor rather than a guess, rounded down so it can only understate. A directory that never shrinks is a directory that has stopped checking.

If you want the raw counts rather than our summary of them, we published a full audit: 400 apply links checked, 70 dead, with the method written out so you can run it yourself.

The catalog's current numbers, the deadlines closing in the next 60 days and the 16 October cleanup are set out with their sources in the State of startup grants, October 2026 report.

Why our numbers are floors, not headlines

Discovery adds programmes weekly and the purge removes them at a similar rate, so the live total genuinely fluctuates. Every public figure we quote is therefore a safe floor below the observed minimum, not the latest reading. If we say 8,000+ grants and 2,600+ India schemes, you should find at least that many, never fewer. The product reads the live count and rounds down. We do not bump a published number because one reading happened to be high.

How eligibility is scored

Four verdicts, and the labels are meant to survive contact with reality. You never see a percentage on the card.

• Strong match. A high fit score and at least one real profile dimension lining up beyond stage. Stage alone is close to meaningless, because nearly every programme open to idea-stage founders matches an idea-stage founder. Treating one stage match as enough is how three generic incubators end up at the top of a SaaS founder's list labelled strong match.
• Worth a look. A good score, or a high score where only the stage lines up. A real programme worth reading, without everything lining up.
• Possible fit. Some part of your profile matches, not enough to rank it higher.
• Long shot. Said out loud, rather than hidden, so the list above it means something.

If you run the full eligibility check on a grant, the verdict changes to one of three words based on what the check found: eligible, mostly eligible, or likely not eligible.

Two rules sit above the score. A rule the scheme publishes and the founder actually fails outranks any score: the grant is marked likely not eligible, whatever the maths said. And a question we never asked is recorded as unknown, never as a failure, so nobody is scored down for a profile field they were not shown. That distinction was a real bug once, and fixing it is why the app now asks rather than assumes.

What we refuse to do

• We do not promise funding. Incubateer shows where your eligibility is strongest so your time goes where it can pay off.
• We never submit an application for you. The AI drafts, you edit, you send. No founder should let a model press submit on a government form.
• We do not inflate a score to look useful. An honest long shot is more valuable than a flattering strong match.
• We do not send a programme reserved for a group you are not in. If a scheme is for women founders or SC/ST founders and you told us you are neither, it stays out of your email.
• We do not sell your data. Profiles and documents are not shared with funders. Your documents are never shown to an incubator, and an incubator sees your profile and applications only after you accept its invitation.

How often things update

• Discovery and the real-source scrapers: weekly, plus a daily maintenance pass.
• Apply-link sweep: rolling, weekly.
• Catalog hygiene, merging duplicates, closing programmes that have ended and expiring dated deadlines: daily.
• Guides and weekly posts: re-read against the portal, with the check date on the page. Anything older than 120 days is not published until it is re-read.

Found a mistake?

Email hello@incubateer.com. A wrong amount, a dead link that survived a sweep, or a verdict that reads as dishonest are all worth telling me about, and I read every message. Corrections go into the guide and the catalog, not into a changelog nobody reads.

Common questions

Where does Incubateer get its grant data?

Four sources. The scheme's own portal, which is where every figure quoted in our writing is read. Curated lists kept by hand, deepest for India and the Gulf, plus listings read directly from the Startup India funding page, the EU Funding and Tenders portal and SBIR.gov. A weekly AI discovery pass that covers India and the Gulf every week and the rest of the world on a slower cycle; its rows are treated as candidates until the link sweep checks them, and a row claiming more than the scheme publishes is dropped. And a weekly browser extractor whose proposals a person reviews. We stopped using grants.gov, the World Bank, NIH RePORTER and UKRI, because nothing they returned was open to an Indian or Gulf startup.

How does Incubateer know a grant is still open?

Apply links are re-checked on a rolling weekly sweep. A link that fails DNS or returns 404 or 410 is struck once, then deleted if it fails a second time. Timeouts, refusals and 403s are never treated as dead, because those are usually geo-blocks or bot defences rather than a closed programme. Grants with a real dated deadline are removed a week after it passes; rolling, annual and twice-yearly schemes are left alone.

How does Incubateer score eligibility?

Four verdicts: strong match, worth a look, possible fit and long shot. After the full eligibility check they become eligible, mostly eligible or likely not eligible. A strong match requires a high fit score and at least one real profile dimension lining up beyond stage, because nearly every idea-stage programme matches an idea-stage founder. A published rule the founder actually fails outranks any score and the grant is marked likely not eligible. A question we never asked the founder is recorded as unknown and can never count against them.

Does Incubateer guarantee funding?

No. Incubateer shows where your eligibility is strongest so your time goes where it can pay off. It does not promise funding, it does not submit applications on your behalf, and it does not inflate a score to look useful.

Why do the grant counts change?

The catalog is live. Discovery adds programmes weekly and the dead-link purge removes them at a similar rate, so the total moves week to week. Published figures are therefore safe floors below the observed minimum rather than the latest reading, so a number we quote can never be higher than what you find in the product.