Answer:
For any real experiment there must be at least five things: an independent variable, a dependent variable, control variables, and two groups often one experimental and one control.
Carlos hypothesis is that by fixing leaks their water consumption would be lower. As such his dependent variable is the water consumption and his independent variable (the one he is manipulating) is the amount of leaks in the house.
In order to make this an experiment Carlos needs to first make sure there is an actual decrease in the water consumed, and second he needs to make sure it's due to the fixed leaks. So to do this Carlos should measure their water consumption during a period of time before fixing the leaks and then measure the water use for the same period of time after fixing. The harder part is making sure that no other variables can interfere, these are called control variables. For example if before fixing the leaks Carlos had a huge party and had to wash a lot of dishes any decrease in water could be due to this. Therefore he would need to either make sure water is used for the same things in the same amounts during these two periods or track any increases due to specific events happening in only one of the periods.