Anything where customers pay to enter and something is allocated by chance sits under scrutiny from three directions at once: the regulator, the card schemes, and customers who will assume the draw is rigged unless you can prove otherwise.
Entry and ticketing mechanics with credit balances. Instant-win allocation with seeded, reproducible outcomes. Free entry routes that carry equal standing to paid ones. Identity, age and phone verification. Automated payouts with failure, retry and recredit handling. Margin and liability reporting per competition. Fulfilment workflows for physical prizes.
Committing to a seed in advance and publishing it afterwards is now standard, and it does prove something real: that you did not change the outcome after entries closed. Most operators stop there.
It leaves the harder question unanswered. A committed seed proves you did not change your mind. It does not prove the seed was random rather than chosen. If the operator generates the seed, the operator can generate a great many of them, run each through the algorithm, and commit to whichever one produces a convenient winner. Every published check still passes. The verification page still says fair.
Grinding seeds until you find the outcome you wanted is indistinguishable from honesty under a scheme that only verifies the seed after the fact.
The fix is to take the unpredictability out of your hands entirely. Commit in advance to deriving the outcome from a public source that does not exist yet and that nobody controls.
The strongest generally available source is a future block hash from a major public blockchain. You publish, before entries open, that the draw will use the hash of a specific future block. That value cannot be known by anyone, including you, at the moment you commit to it. It cannot be influenced without an attack on the chain itself that would cost more than any prize you are offering. And it is permanently public, so anyone can check it years later.
Combine that with your own committed seed and neither side can determine the outcome alone. You cannot grind, because half the input does not exist when you commit. The chain cannot target you, because it has no knowledge of your competition. Verification becomes an arithmetic exercise a customer can do with a block explorer and a hash function.
The practical detail that matters: choose the block far enough ahead that it is not already close to being mined at the point of commitment, and publish exactly which chain, which block height and which derivation you will use. Vagueness at that step reintroduces the discretion you were removing.
The free route is load-bearing. Where an exemption depends on a genuine no-purchase entry, that path has to be real, equally weighted and equally easy to find. Treating it as a legal footnote implemented badly is how an exemption stops applying.
Liability is live, not periodic. Every entry sold changes your exposure and your cash requirement at draw. That number needs to be visible continuously rather than assembled at close, because it drives whether you have funds in place to pay out, how you price the next competition, and whether one currently running needs more marketing while there is still time to do anything about it.
Acquirers treat the category as elevated risk. Onboarding is slower, pricing is higher, monitoring is closer, and single-provider dependency is a standing exposure rather than a theoretical one.
Credits are money and behave like it. The moment a balance can be topped up, spent, won, gifted or refunded, you need a ledger rather than a column. Winnings, bonuses and refunds all move between states and every movement needs a record.
Verification sits before the payout, not before the entry. Front-loading identity checks kills conversion. Deferring them until a payout is due means the check happens when the customer is most motivated to complete it.
We have built real-money entry platforms end to end, including allocation, verification, payouts and the reporting that tells you what you actually owe.