Disappointed in device discovery


#1

I’ve had my Sense installed for a month and compared to recent reports I’ve read here, I’m simply not seeing device discovery the way I would have expected. At this point, all it has found is one of two refrigerators, the water heater and an unknown random device that seems to only power up on Monday. I seem to be well behind in discovery vs many other recent reports I’ve read. The fridge and water heater were found within the first week and nothing since. I saw the earlier reports from 2016/17 when discovery was very slow but recent reports seemed to indicate I could expect more rapid discovery.

At this point, I’m pretty disappointed that things like the air handler for the A/C which is on and off countless times a day or other appliances like washer/dryer, dishwasher or kitchen fridge aren’t being matched. The only devices it has found are within physical arms length of the sensor in the garage. All show up in the timeline view in real time when they power on and off but none are being identified to allow me to learn their power usage over time which was the primary reason for purchasing and installing the Sense.

I know it will never find the outside compressor for the heat pump and the pool pump due to the configuration of my power panels in the house but I had expected it to find more of the devices inside the house.


#2

Thanks for writing in, Don. That does sound like pretty minimal detection. Have you worked with Support on this at all? They can look at your data and diagnose some potential blockers, like a poor wifi connection and so on. That said, device detection can just suffer in some homes due to noisy power and other related factors. I do suggest reaching out to support as they may be able to offer some insight.


#3

After a year and a half, Sense has only found about a dozen devices (and some of those not reliably) and in my first month it found only 1. Apparently Sense’s abilities are HIGHLY dependent on your specific devices and even more on your environment. In my case, Sense engineers have identified one very distinct signature (my constant pressure well pump) that interferes with all other device detection, and isn’t itself detectable.

So, you aren’t doing that badly.


#4

Thanks. I wanted to verify that my expectations were reasonable before opening a support case which is now started. I’ll share here if I learn anything that changes the pattern.


#5

I am an early adopter, and there are many many devices that have never been detected on my system. Could you take a look?


#6

It looks like my expectations were unreasonable after all. I just got a response back from support that was essentially a copy and paste from the web page FAQ of ‘it takes time’ and you just ‘have to be patient’. Kind of disappointing that after the product has been on the market this long, this is still the response. Clearly the marketing message is still way out ahead of the actual capability of the product.


#7

Definitely huge promises and, for many of us, really poor delivery. I’ve also talked to folks who love it, so for some portion of their customer base it is working. After a year and a half, I have given up. Their engineering basically told me “we can’t handle your power signatures”.


#8

It’s clear at least a subset of Sense users have had similar experiences: the reality does not come close to the marketing message. In my case my three biggest loads (pool pump and two EVs) have yet to be found after 18 months.

I would hope Sense would tone down the hype as a result of this feedback, and yet the main message on the web site is “See what’s up. Know what’s on.” Based on my experience this is/was false advertising.


#9

Yup, indeed!


#10

Dear Andy,
You grumble a lot.
Your BFF,
MachoDrone

MachoDrone bought this with low expectations knowing it is the right price and if it ended up only allowing me to know my whole house kwh each day, I’d be very pleased… That’s all any other genetic device would do prior to Sense, which I used, and it was beneficial.
I will admit I’m pleasantly pleased with my results, but I set my expectations low on everything, so I can see a silver lining.

In all seriousness, do this, just a weird idea @RyanAtSense may be able to shoot down or agree with… only place your clamps on the well pump leads to it’s breaker, after that’s found, then move the clamps to another 220V breaker lead, like your A/C… Once that’s found do another 220V device, like water heater, etc… What imma trying to do is just see if once sense finds these devices, will it separate the signatures from 120V things… If so, cool, but this is just a dumb I idea I would try… Maybe go opposite direction and do 120V lines first, but the CT clamps always need to be on the corresponding legs of the mains… If this doesn’t make sense then it’s not for you.
I would do it… However I’m MachoDrone. If I grumbled about a device that is working great for many users, I’d find a way to overcome the issue. I hope it works. Lemme know.
P.S., also Andy, please click the heart in my post.
… … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … V heart V


#11

Yup, I expect things I purchase to work properly, as the manufacturer promises. I already had a whole house system (www.welserver.com) for nine years, and it does an excellent job of monitoring total use, solar generation, and on/off for HVAC, water heater, ERV, etc, plus temperatures. So, I bought Sense based on their promises that it would provide me data on individual users…at last count my home has over a hundred.


#12

I think I’m doubly frustrated by the feeling that I just got blown off by support with a copy and paste response. I started out as a big proponent of this solution when I first installed it but based on my disappointing experience with device discovery, I’m not likely to recommend to friends. If I do, they’ll end up coming to me asking why I suggested this if their experience in device discovery is as disappointing as mine.

I’m just not a fan of living with a product that has a nebulous commitment of eventually it ‘might’ become useful, or it may not work at all in your home. I guess I shouldn’t have watched that episode of This Old House- it kind of hyped it well beyond what it can actually do.


#13

Yup, me either and I’ve engaged with Ross Trethewey about that exact problem via e-mail. He tells me:

· His experience is quite different, and Sense is very good in his home

· The whole intent of that part of the show is to highlight bleeding edge home technologies

· They don’t cross check the various technologies

I pointed out to him that TOH has become a trusted advisor over many decades to large numbers of us and that when TOH says something we tend to take it as accurate. I also pointed out that by hyping things that aren’t proven, they cast serious doubt on the TOH brand. I urged that they be up front and totally honest about the status of any such products they highlight.

Moderator Edit: removed bolded text


#14

It’s unsettling to hear about your experience with Support. We strive for excellent Support and have gotten great feedback on it in the past. I’m definitely going to pass on your feedback to our head of Customer Service. I’m guessing (only guessing) that your support agent verified that everything was working correctly behind the scenes before responding, but I’m going to reach out and make sure.

For everyone – device detection does vary pretty widely across homes. As noted above, we have users with just fantastic detection (I wish I was one of them!), and users with pretty terrible detection even after a year. It varies based on a lot of factors that we’re still working on pinpointing and then hopefully figuring out a way to mitigate. We are working on being more transparent about that in our messaging and I was brought on partially to better manage user expectations. We definitely need to do better work on that front, I wholly admit.

I do like to suggest just waiting it out, but I understand that this isn’t palatable by everybody. We offer a 60-day return period for this reason. Sense is still very new tech (at least the machine learning aspects) and we’re working constantly to improve it. We have a growing data science team who are 100% dedicated to making device detection better and have seen improvements even in my short tenure here at Sense. We wish we had a magic button, trust me. If you are totally unsatisfied to that point that you just no longer want Sense in your home, reach out to me and we’ll do what we can to make it right.

Regarding ToH, we had no input at all into their messaging, but I’ll keep working to refine ours.


#15

FWIW, my discovery experience has been pleasing to me: about 20 devices in 6 months, including distinctly identifying a second microwave 1-2 weeks after we bought and started using it, the main fridge vs. the wine fridge, the hot tub pump and heater (which I’ve found is, by itself, over 25% of my total usage!!). I really only bought this for realtime analysis, so aggregating usage per device over time is nice. Sense only detects about 4 out of 50 light switches in the house, but I don’t expect better.

I wonder what accounts for a good learning experience versus a bad one. Certainly usage patterns: it’s mostly the things that turn on and off frequently regularly that have been identified. But is there something about the wiring in my house, or the placement of my Sense, compared to others that might account for the discrepancy?


#16

LED lights are a major challenge. They’re low wattage, which can already make for tough detection, but their signatures are pretty noisy as well.

As for your last question, the big thing is line noise. The noisier your home power is, the tougher time device detection will have. It makes perfect sense if you think about the resolution we’re gathering data (1 million times every second). At that sampling rate, even otherwise minor interference starts to look pretty major. But yes, usage patterns matter too. If you’re not using devices frequently, Sense won’t get the data it needs. Similarly, if you run devices 24/7, Sense won’t be able to detect them because it will never get a chance to recognize any on/off transient patterns. And then there’s Wi-Fi. Sense is absolutely reliant on a solid Wi-Fi connection for device detection. If your signal that the panel is poor, detection can suffer immensely. There’s more to it than that, but those are definitely the big ones.


#17

Is there anything one could install in the breaker panel to reduce noise on the lines and help w/ detection?


#18

Yes indeed, and my very poor detection history was dominated by my “extremely noisy” Franklin Electric constant pressure deep well pump. To a human being, the pattern is very distinctive and trivial to recognize, but to the Sense algorithms it’s just HUGE amounts of noise. So, whenever the pump runs (domestic water, geothermal heating in winter and geothermal cooling in the summer), it apparently makes it impossible for Sense to detect new devices and to accurately monitor already detected ones.

Perhaps Sense needs to be up front about the kinds of devices that will render the Sense device ineffective?


#19

So, how does one know if the Sense platform sees their house as ‘noisy’ and therefore it isn’t likely to ever work as advertised in their environment? I had the hardware sitting on the shelf for many months following purchase before I installed it so I’m now well outside any 60-day return window. My biggest concern now is that I’m being told to wait multiple ‘additional’ months to see if it ‘might’ work in my house, or what I have now might be all I’ll ever get which is not the solution I thought I was purchasing nor what the website and other messaging in the market suggests. If what I’m reading about is that LED lights introduce ‘noise’ and my house is 50% LED now is this ever going to work? I’m certainly not taking those out as they are a big part of my energy use reduction strategy.

Also, I’m sorry but you can’t distance yourselves from the ToH segments since Sense staff actively participated in at least the first one and it’s even referenced on your own website and in marketing emails you’ve distributed. And if the message was inaccurate (which I believe it was), you have a responsibility to push them to correct it in the second segment when they included your product on the show again. You can’t just take the message that can result in sales and not own responsibility for the product not living up to the message your own people presented in that segment.


#20

Totally agree, and your president and head of research (both of whom I’ve talked to about the problems) were featured on the show. I’ve also been discussing this with Ross Trethewy (sp?), pointing out that ToH has been a trusted advisor to millions of us over the decades, and it’s NOT a good idea to be promoting things that don’t actually work for some (significant) portion of the customers.

BTW, Ross does say it works great in his home, so he’s one of the lucky ones.