There is a 'payout ratio'. How often people win is carefully calculated.
Interestingly, just like Vegas, this has been EXTENSIVELY researched, with hundreds of thousands of data points (from weekly collection data).
People VALUE something more if they win it.
Wal-Mart sells a plush toy for $3.00. The teddy bear crane at the front of the store has almost exactly the same product. People will be FAR happier when they WIN the toy instead of purchasing the toy.
So here is the math: For a toy that people value at $3.00, if you let them win too often, the crane won't get played as much. They think 'I can win a toy anytime I want' and the crane won't make as much money through the coin slot... and there is a cost of toys going out of the crane, but mostly it's about getting as many plays as possible from people.
This is counter-intuitive, but is ABSOLUTELY true, proven with a lot of data. When people win more frequently than the 'optimal point', they don't revisit the crane as often.
Similarly, if people never win, they'll engage a certain amount of times, but will quickly decide to not play crane games anymore, or at least stop playing the crane that doesn't meet their expectations of being able to win.
It is FAR worse to not allow people to win toys than to allow people to win too often, but both ways you can set up the crane to make you less money.
The sweet spot for a toy that people value at $3.00, is exactly fifteen plays. At 50 cents a game, that's $7.50 in the cash box for a toy valued at $3.00.
At a dollar a play, the numbers are different. We are usually dealing with a perceived value of the toy of $6.00 or more, which changes the equation. I haven't been involved with this for a while, and right off the top of my head I don't know the exact payout ratio is for dollar cranes.
The point is this. People value things they win more than they value things they purchase. Smart crane operators will make sure you win exactly often enough to keep you coming back.
'Intermittent reward is the strongest motivator'.