I would suspect that:
1. It takes a bit of effort to implement
2. In general, people with stuck transactions are like that because they don't want to pay high fees, so aren't likely to part with large sums of cash to push a transaction through.
3. If you had that sort of money to spend on getting a transaction sent, you'd just pay a high fee to start with

So, personally, I don't think it would be much of a money maker at all... sure, you'd get the odd customer here and there... but making ~$300 on an accelerated transaction when you get US$350K+ per block is nothing... I'm not sure the ROI on trying to implement the payment/queueing system and messing with the pools mempool/transaction selection algorithm is really worth it.