IIRC this falls into classic fallacy where teams set low goals so as to always exceed them. The better way is to have people estimate and bet money on delivery.
Maybe a bigger company could do it like some card games decide on who will choose the game: Make several teams "bid" on the contract, the one with the best bid gets it, but winning the game (e.g. getting bonuses) depends on if they can actually deliver on their bid.
