Why a cool-down beats a rating
The efficiency figure on a water heater label is produced by a standardised test with a defined pattern of draws, and standby loss is already inside it. That is convenient for comparing two new appliances on a showroom floor and close to useless for working out what your own tank leaks, because your house does not draw water the way the test does and your tank does not stand in the room the test used.
A cool-down needs nothing but a thermometer and a night when nobody runs a tap. Read the tank, wait, read it again. With the defaults here — 40 gallons, 140 F falling to 132 F over twelve hours in a 65 F room — that is 333 pounds of water giving up 2,666 BTU, or 222 BTU an hour. Divide by the 71 degree difference between the average tank temperature and the room and you have 3.13 BTU an hour per degree, which is the number that belongs to the tank itself and can be carried to any other room temperature.
The hours are what make it money
222 BTU an hour sounds like nothing. It is roughly a nightlight. But it runs 8,760 hours a year, and at a 75 degree difference year round that comes to just over two million BTU of heat, which at 60 percent into the water means about 34 therms of gas. At a dollar fifty a therm that is 51 dollars a year to keep water hot that nobody touched.
Sixty percent is a placeholder in that field, not a claim. Put in whatever your own appliance states or whatever you can work back from a bill, because the fuel figure scales directly with it: at 90 percent the same measured loss costs 34 dollars a year instead of 51.
Reading the result honestly
The single biggest error in a cool-down test is the tank cycling in the middle of the window. If the burner or the element fires, the second reading is not the bottom of a cool-down and the loss comes out far too low. The second biggest is a draw nobody noticed — an ice maker filling, a dishwasher on a delay start, someone washing their hands at two in the morning. Both faults push the answer the same way, which is why a result that looks implausibly good is usually a spoiled window rather than a good tank.
Room temperature is the other lever, and it is not a nuisance variable. A tank in an unheated garage genuinely loses more in winter, and the yearly figure this page prints assumes the room stays where you put it. Measuring once in each season and averaging is more work and a much better number.
Where the comparison stops
When you enter a standby loss and a price for a replacement, the page compares standby against standby and nothing else. That is deliberate. The energy in the hot water you actually use is heated by whichever appliance is there, at broadly the efficiency you entered for it, and folding it in would bury the one thing this measurement is good at. It also means the payback figure is a floor rather than a full case: a tank that is leaking water, a control that has failed, or a household that runs out of hot water every evening are all reasons that have nothing to do with the number above.
The comparison can come out negative, and it does so most often when the replacement is a bigger tank. A larger surface loses more standing heat, so a 50 gallon unit can measure worse on standby than the 40 gallon one it replaces while being the right appliance for the household for reasons this page cannot see.
Questions people ask
How long does the tank have to sit before I take the second reading?
Long enough for the drop to be bigger than the error in your thermometer, which usually means overnight. A one degree drop read on a thermometer good to one degree tells you nothing. Eight to twelve hours with the taps shut gives a drop of several degrees on most tanks, and the arithmetic divides by the hours anyway, so a longer window is a better window as long as nothing draws water.
The temperature barely moved. Is that good?
It might be, or the heat cycled on during the window and refilled the tank behind your back. A cool-down where the burner or the element fires is not a cool-down. Check the drop against how long you waited: a very small drop over a long window with an obviously warm flue or a warm element circuit is the signature of a spoiled test rather than a very good tank.
Why does the room temperature matter so much?
Because heat leaves the tank in proportion to how much hotter it is than what surrounds it. The page reads a loss per degree from your window, then scales it to the difference between your thermostat setting and your room for the yearly figure. Move the same tank from a 70 F utility room to a 45 F garage and the standing loss goes up by about a third at the same setting.
Does this tell me whether to replace the water heater?
No, and it is not trying to. It prices one specific thing — the heat a tank loses standing still — and compares it against a standby figure you read off the label of something else. A failed control, a weeping tank, a household that has outgrown the capacity, or a fuel change are all outside it. Take the number to whoever is quoting the work and let it be one input among several.
Can I use this on a tankless heater?
Not usefully. There is no stored volume to cool down, so there is no drop to measure and the whole method has nothing to work on. The comparison field will still accept a standby figure off a label if you want to price one against a tank you have measured, but the measurement half of this page is a storage tank test.