Short Answer: Your results should be much closer. Several people have done deep analysis and found Sense to be within +/- 2% of their “revenue meter” (metering by the power company).
Long Answer: Your results should be much closer. But discrepancies do happen, for a number of reasons documented throughout this forum.
- An unclosed or angled CT on one of your incoming power lines (mains).
- Sense installed on a sub panel rather than the same house mains the “revenue meter” is attached to
- Networking issues that prevent Sense from transmitting full data back to the mothership. The Sense probe can often backfill if the the network goes offline for a while, but not always.
- Other associated Sense probe issues plus uploads of new firmware to the probe
One way to figure out what’s causing the differences is to do an hour by hour comparison for the month, if your utility allows you to download that. Sense data can be dowloaded and filtered on an hourly or daily basis for comparison. Your discrepancy is big enough that day-by-day comparisons might also point to the smoking gun.
My detailed analysis here: