This isn’t specified. I would assume it’s a unix epoch timestamp, except that a typical row from our database is sent a timestamp like ‘100239866’, which represents March 1973. We weren’t recording data then.
I’m assuming you’re asking this question in regards to Raw Data. It is indeed a standard UNIX timestamp (I’ll make a note to our docs team to be more specific on this point), but note that there are two date fields, and it’s possible you’re looking at the wrong one for your purposes. The first field (identified as ‘ts’ in our custom headers) represents device time. It’s possible for that to be affected by the device’s own clock, which might explain the March 1973 number. The other value, identified in out headers as ‘submit_time’ is the server time when the event was received, and as such is not subject to device shenanigans.
So my initial guess is that the 1973 numbers represent a device with a zeroed-out clock. If you think that’s not the case (for example, if you’re seeing that timestamp when looking at data from a device you know to have a properly-configured clock), it might be appropriate to open a support ticket for further investigation.
Hope that helps!
Thanks for the help!