The first way is to increase the average stakes being played, or the average tourney buyins paid. This doesn't lend itself to rigging, since it is handled very well though the legitimate efforts of marketing, player retention and education, and varied and interesting game structures. I don't think this method is feasible, but maybe someone thinks of a way. I'll mention for the short bus riders that making tourneys play faster doesn't qualify because that is very simply done by changing the structure, it doesn't need to be rigged.

The other way, and the one that any rigged scheme really has to accomplish over time, is to have more butts in seats (virtual ones) on a 24/7 averaged basis. Actual numbers can be found on pokerscout, giving you something to work with. Revenue is a direct function of the 24/7-averaged cash players in seats, multiplied by the stakes multiplied by the rake formula, plus average tourney players in seats multiplied by the average entry.

More players in seats can be accomplished by making bankrolls last longer, or by getting players to deposit more, or by getting players to not quit the site (and obviously by getting new ones, but that is marketing). And maybe you can think of other ways, like a new deposit boomswitch or something, or a cashout curse. The most obvious way is to make win/loss rates bunch closer together for most players so the bankrolls lasts until the money is all churned into rake. Let me say right here, simply making pots bigger will take a very convincing math explanation of how that will increase site revenue. It increases rake for one pot, it does not make money churn longer between players. That doesn't mean pot sizes can't be part of your scheme, but keep in mind you'll have to show the math.

To qualify for the prize, the scheme must be capable of adding 5% to the site revenue, assuming it has been in place long enough to stabilize. This must be backed up by math, including pertinent details like average pot sizes and stakes, rake, occupied seats, hands per hour, etc., and the effect of the scheme on those things. My opinion is that anything less than 5% revenue boost could never justify the risks at a large site. You will need to use realistic numbers in your math. Assume a billion dollar site to make it easy, and add $50 million a year. Adding 10% or $100 million doubles the prize

Personal databases have all of your holecards. Observed databases only have community cards and showdown cards. So different tests can be used on each type of database. One has a complete set of holecards, and the other has enormous sample sizes and multiple player histories. The scheme needs to avoid detection in both. It might also be possible to design a scheme that is only detectable if players pool their personal hand histories. Avoiding that one would be a bonus.

You must assume that at some point the available data will be examined by statistical experts who know how to test the data. Assuming no one will find something sitting there to be found is not acceptable. So it has to be difficult for experts to prove something is manipulated. The forum consensus on this will determine if the requirement is met for the prize.