Where is the rest of the wattage? Alway On


I’ve been playing around with Always On and several other things lately and ran into something I don’t understand.
When my fridge and freezers have been on lately, the wattage is not accounted for.
Both the fridge and freezer are detections that I thought were very accurate about reporting. But this morning both of them were on and they showed off on the device page. Also on the device page, there was nothing for “other”. They use about 200w together.
So where does that wattage go?

I did find that in this mornings instance that the missing wattage went into “Always On”

After I turned off both appliances for awhile my ALWAYS ON went from 561 to 277.
I had turned off some other smaller thing too.
If you happen to have AO numbers that you believe are higher than they should be then this is probably happening with you.
When your usage is at its near or absolute lowest and anything that isn’t always on happens to be on at the time, it will contribute to your AO calculation.

1 Like

The wattage for those devices does not “go” anywhere within this instance. The exact way that it’s determined is a little bit more complex but the jist would be that Sense always performs a check that the total wattage reported by devices plus the Always On value is never greater than the total wattage reported by the Sense monitor at a given time. There’s some fuzzy leeway, but this is the basic principle that Sense should follow.

If a device turns on which Sense has detected but the reported wattage for that device plus the current Always On value would cause for the total reported device usage to be greater than the total overall usage, then Sense will not allow for that device to show as being on. The Always On value itself will not be influenced in any way by the change in overall wattage usage from those devices.

The Other bubble is just a value that represents wattage used which Sense cannot attribute to a known device. So if your total usage is lower than the Always On there will be no Other bubble.

While it shouldn’t ever really happen that the Always On is greater than the total wattage – as the Always On bubble is the only non-real time bubble – it’s certainly possible to hit a number of situations where it can be. Those are the basic, basics. There’s a lot more to it than that which goes on behind the scenes, but it’s a bit complicated beyond that point.

I’m not on the forums too much and usually one of the better people to ask for these types of questions. If there’s ever any questions, comments, or observations that you (or any one else reading!) ever have in regards to how Sense is capturing or calculating things, it’s typically best to write into support directly to me. It may not go straight to me, but as long as you mention Tyler, someone will pass it along and I’m usually on point with responding. Except Fridays, because who works Fridays? (I do, I just work on physical monitors on that day and don’t check my e-mails much.)


If it won’t allow the device to show because the total with the Always On makes it too high, then it’s sounds like it is “going” somewhere. Like Sense realizes “big mistake, this isn’t possible, better hide this extra 200watts” If the sum of everything put it over like that, wouldn’t it be much more accurate to reduce the Always On instead of hiding or not showing those devices? I’m talking specifically about the fridge and freezer situation from this morning.
Doesn’t this skew or otherwise make usage for those hidden devices inaccurate?

Always great to see a user with @sense on here. Thank you


I have the same issue. My new VS pool pump runs 2 hours at 2400 RPMs and 8 hours at 1300 RPMs. The 8 hour run is totally consumed in Always On. A portion of the 2 hour run is also consumed in AO, but I do have some Other during that run.


If you can set the pump on a timer then you could manipulate and pull it out of always on.
I did more experiments and this morning mine now reads 184w, it was over 900 a couple weeks ago and almost 600 a little I’ve a day ago


The pump is on a timer - a 24 hour clock. There are 4 programs that get set based on the 24 hour clock.

At low speed the pump usage is not reported on Other. At higher speeds, part of the pump usage is in Other. I’ve played with this as well. There’s a “reserve” in AO that get’s filled by undiscovered things, Other.

1 Like

If you can locate on your timeline the quietest part of the day when your usage is lowest, try this.
At that point make sure everything you can have turned off is off, especially the pump. Make sure it’s like this for at least 15 minutes and within a couple hours, your AO should change.
I found where to do this in my house by using the power meter in the app under usage/trends.


Thanks, I’ll give this a try.

1 Like

My 2c - Always On is a statistical model, not an exact thing. If you have lots of undetected or partially detected devices that are on and off all day and night (fridges are good examples), it’s very possible that your Always On doesn’t really reflect the small subset of devices that are really “always on”, but mostly the statistics of a bunch of Poisson Processes (a series of discrete on/offs where the average time between different events is known, but the exact timing of events is somewhat random). Once that happens Always One is really a statistical thing that can’t be linked back exactly to specific devices so correction to keep Always On from being bigger than Total Usage, really don’t come out of “somewhere”.


I agree that it’s not accurate, it’s more of a rough estimate of what “could be” Always On. I understand that we really can’t directly attribute some of this wattage in always on directly to a specific device but there is a link there.
If my lower use is 200 watts and at that time the freezer is on, that freezer just became part of the calculation. Later in the day the lowest may again be 200 watts but the fridge is on and the freezer is off. Now the fridge is part of the calculation.
The way Tyler explains it, if the total of always on and all devices exceeds what is being drawn at that point in time, Sense hides some of the devices to compensate for the overage.
What I’m saying is instead of hiding a device that sense knows is on, reduce the Always On accordingly.
I can see that in my home, even when we are at our lowest use there are still things “on” that are not “always on” and is most likely the case with everyone.
I believe the little tests I’ve done over the last couple days proves that this needs more work.
One of the selling points is helping identify “vampire” devices and where our energy is being used.


When sense decided to hide my fridge and freezer from the device page then where did that data go?


Guessing it became part of Always On…


I honestly don’t know what happened to it. I thought that since it was running but Sense wasn’t showing it on the device page at that time because it put the total over, maybe it still was on the Power meter for the individual devices and the usage stats for those devices.
I went back and looked at the time when all this was happening, checked the fridge and freezer power meters and it appears they are both missing a cycle when compared to normal use.
Does it make sense that I think if a correction needs to be made because everything running and always on don’t add up that it shouldn’t be to hide devices known to be on but to reduce Always On since it’s not a live reading anyway?
I don’t understand why the Always On would be the absolute


Nothing becomes part of anything. It’s a sanity check that’s applied to any device that Sense thinks has turned on that exists for a variety of reasons. As a very simple explanation for how this check works:

Any time any device turns on the device model says “I expected this device to be using X watts”, Sense replies that the house is currently using Y watts in total with Z of it being reported by other devices. Provided that X + Z < Y, then Sense will show that device as being on. If X + Z > Y then Sense says “No, this model must be wrong, something else came on instead.”

Again, that’s a super simple explanation. The actual checks that are performed are a good deal more complex than that, but this is the basic principle behind the wattage check. If the Always On is over-reporting then it’s going to block Sense from saying that devices turned on until the total wattage in use is greater than the Always On + anything else that’s also running.

Sense does not put this power into your Always On. The Always On value is not going to change, it is not going to grow, and the power from those devices was not previously being tabulated into the Always On.

In 99% of user cases, the Always On value should never be over reporting. This will only happen if there are devices which are left on for 20+ hours which then get calculated into your Always On value and then you turn those devices off. Until the next period wherein the Always On recalculates the value for Always On will be higher than it should. It’s super rare for this to ever be significantly higher than what the actual Always On value should be.

So, yes, if you forget that you left your basement lights on which use 200 watts and then turn them off a day later, they are going to get converted into Always On value for a time until after you’ve shut them off. During that time period, it’s probably likely for Sense to miss some things like your fridge turning on, but even that is more rare because there’s some fuzzy logic in how Sense allows for over-reporting from devices so if there’s some other light in the house that’s using 120 watts that’s on, then your fridge will probably report fine because Sense won’t care that the total device usage is 80 watts or so off from the overall usage.

Again, that’s simplified. The fuzzy leeway that Sense allows isn’t any set specific value, it depends on a lot of other factors. And none of this even factors in smart plugs which are a completely different thing.

Regardless of all this, the Always On value for a home should be relatively stable and consistent. If you are noticing that your Always On in changing by high values throughout the day or day to day; certainly check with our support team to make sure there isn’t a bug in the app or in the detection of something, but outside of a bug it’s caused much more by how you are using energy. There’s something that’s being left on for far longer than it probably should be, and I’d certainly consider that a vampire load.


I don’t know specifically because I haven’t looked, but I’d assume from the way you describe it, the pump runs 24/7 but it just has different speeds at different times? It never actually shuts off for any period of time, just has a period of high usage and a period of low usage?

If the pump only runs for 10 hours a day, then this issue shouldn’t occur, but if the pump just runs through cycles of 2 hours on high speed then 8 hours of low speed then directly back to 2 hours of high speed, then, yes, the low speed power is going to be part of the Always On.

The pump would always be using that much energy, it would only be during the higher speed time where the energy increases that it would ever use more power but the base power that it used during the low speed portion would still be there.

If the pump is only running for 10 hours a day and is reporting like this, then that’s something I can look at if you write in.


Thanks for the response. The pump starts at 9 am and runs for 2 hours at high speed. After 2 hours to runs for 8 hours at low speed. Upon completion the pump turns off and stays off until 9 am.


From 180 to 530 today. My always on hasn’t updated from last Thursday until today. I knew it should go up but incrementally and not that sudden and dramatic. It shouldn’t take nearly a week for it to update.


I posted a while back about something like tihs happening to me. the added usages in the bubbles were greater than the amount reported as the total. or something like that. I only noticed because it was saying I had 2 air conditioners (i don’t) and the discrepancy was huge.