The ROI of game purchases are often portrayed as the cost of the game divided by gross sales.  Of course, the gross sales vary greatly from location to location.  Almost everyone takes that into account when considering game purchases.  However, there are three other variables that should be considered and are often overlooked.
1) The Payout: There is a wide range in the amount of tickets different games give out.  The value of those tickets needs to be subtracted from the sales of the game to get a payout number.  Most locations do not accurately calculate their ticket value because there are usually multiple variables to get a true ticket value.  Most of the highest earning games, also payout the highest, so this is important to measure.
2) Target Demographic: When a new game is added, some of the sales the game generates will be cannibalized from other games.  This is impossible to measure exactly, however you can consider the demographic a game targets which informs the degree of cannibalization.  For example, games targeted to small children usually don’t perform well.  However, if there are no games targeting the 5 and under group, and one is added.  Most of those sales, are sales you wouldn’t have had, so those sales are more valuable than the average game.  Similarly, a location with a lot of corporate events may have an abundance of basketball games.  If having 4 basketballs is appealing to a party planner and get the location a $10,000 corporate event, they are well on the way to paying for the extra basketball games.
3) Time Period Measured: Some games stay popular longer.  A popular video game will typically have a 6 month to 1 year life before sales drop significantly but will earn very well.  While a Skeeball will earn for years on end, as long as it is maintained, but will earn significantly less the a hot video game.  If we compare the ROI on say a Jurassic Park video game vs a skeeball for 2017, the Jurassic Park will be much better.  However, if we measure the ROI on the same two games over the next 5 years, the Skeeball will probably win, and over 10 years it will certainly win.
This business is not rocket science but it is science.  It is important to look beyond the surface of game sales to really maximize profitability.  This is just one area of arcade management that is often oversimplified.