We re-read all 186 entries in our own catalogue once a week, by hand, because the alternative is worse than not having a list at all. Last month that process caught an expired domain now serving a casino affiliate site, an accelerator that quietly turned itself into a paid course, and a government grant agency that no longer exists. None of those three looked dead from a search engine's point of view. All three still ranked. A founder who found any of them on a generic 'top 50 startup grants' post would have spent an afternoon writing an application into a mailbox nobody reads.
The five checks that take two minutes each
Before you write a word of an application, run these. Each one takes about two minutes, and together they catch almost everything that kills an application silently.
- Last cohort date. If a programme's most recent named cohort is more than 12 months old and there's no announcement of a next one, treat the page as archived until proven otherwise.
- Does the application form actually load. Not the landing page, the form itself. NLnet Foundation Grants publishes a hard next deadline (November 3rd 2026, 12:00 CET) precisely because its form opens and closes on a cycle, which is the kind of detail a dead listing never bothers to carry.
- Has anyone posted about it in the last year. Check the organiser's own social accounts and, where it exists, the programme's own grants or developer forum. Silence for a year is a stronger signal than anything on the programme's homepage, because homepages are the last thing anyone updates.
- Does the organisation still exist. Not the programme, the organisation behind it. A foundation can dissolve while its old grants page stays live on a server nobody is paying attention to.
- Does the domain still belong to them. Expired domains get bought by whoever will pay for the backlink value, and a startup grant page is exactly the kind of high-intent traffic a casino affiliate wants to inherit.
What each failure mode actually looks like from the outside
The expired-domain case is the easiest to describe and the hardest to catch with a glance: the page design looks slightly off, the grant copy is still there, but the links in the footer go to unrelated betting sites. Google doesn't de-index it fast because the domain's old authority is still intact. The 'became a paid course' failure is quieter. The programme used to disburse capital or credits; now the same brand sells a cohort-based course with an application form that looks identical to the old grant application, down to the field labels. You only catch it by reading the fine print on what you're actually being asked to pay for. The 'organisation no longer exists' failure is the one aggregated lists are worst at catching, because the page itself never changes. Nobody took it down. It just sits there, technically true in 2021 and false now.
Why a list of programmes rots faster than the programmes themselves
A programme changes its own terms when its treasury runs low or its mandate shifts. That happens on the organiser's own schedule, and good programmes are transparent about it. The Graph grants program, for example, currently states plainly that there are no open RFPs at this time, rather than leaving a stale 'apply now' button up. That's a live programme being honest about a dead window. A list, by contrast, has no equivalent internal pressure to update itself. Nobody's treasury runs out when a blog post goes stale. The post keeps earning search traffic and ad impressions whether or not the $1M pool it describes, like Trailblazer 2.0's $1M pot with a $10k cap per submission on Arbitrum, has already been fully allocated. The programme has a reason to correct itself. The list doesn't, which is exactly why a list needs someone re-reading all 186 entries on a schedule instead of writing them once and walking away.
What a maintained listing should actually show you
A listing worth trusting states its uncertainty instead of papering over it. Our own entries for Colorado OEDIT's Advanced Industries Accelerator Grants and the Business Chain MVP Grant are both flagged 'status not confirmed' rather than quietly left looking current, because the first's site couldn't be checked automatically and the second doesn't publish current application dates. That flag is more useful than false confidence. Compare that to entries we can confirm in full: ElevenLabs Grants Program offers more than $5,500+ in value on a rolling basis with no deadline, Starknet Seed Grants go up to $25,000 in STRK reviewed in ongoing rounds, Base Builder Grants pay out 1 to 5 ETH per award, and Together AI's Startup Accelerator scales from up to $15K in platform credits at its Build tier to up to $30K at Scale and up to $50K at Growth. A maintained listing names the exact sum, the exact deadline or cadence, and tells you outright when it couldn't verify something, rather than filling the gap with confident-sounding filler.
Run the five checks yourself even on a list you trust. We re-check 186 entries a week and we still find things like Conflux's build grant or Fuse's Web3 Business Grant Program changing their disbursement terms between our visits, because milestone-based programmes like Flare's Grants Program adjust their staged payouts without a press release. Two minutes per check, five checks, before you write a single word of an application. That's the whole discipline.