Different Measure Values between Desktop and Service

[Anonymous]

I have the following measure to get the previous month's Quantity based on yesterday's date. 

PreviousMoM = CALCULATE(SUM(factSales[Quantity]),filter('Date','Date'[Date]>=EOMONTH(TODAY()-1,-2)+1 && 'Date'[Date]<=if(DATE(YEAR(today()-1),MONTH(today()-1)-1,DAY(today()-1)) > EOMONTH(today()-1,-1), EOMONTH(today()-1,-1), DATE(YEAR(today()-1),MONTH(today()-1)-1,DAY(today()-1)))))

Something weird seems to happen toward the beginning of the month (I noticed it yesterday on the 3rd in the afternoon, but today the 4th everything is back to working fine). Yesterday after the data refresh, I noticed that this calculation was returning Previous MoM values for Feb 1-3. The expected result was Feb 1-2, because it's as of the previous day. 

Now the main problem seems to be between Desktop and Service. When I refreshed the data in Desktop on my local machine, the expected values were returned. Only the version that is published to Power BI Service was including the Feb 3rd data. I republished just to be sure, and the exact same pbix was correct on my Desktop and still incorrect on Service. Since it's correct today, the 4th, I'm wondering if there's some system date difference on Power BI Service. The machine that the gateway is installed on is US Mountain time, my local computer is US Central time, and the refresh schedule is set to use US central time.

Can somebody help me figure out what is causing this difference?

[v-frfei-msft]

Hi @Anonymous ,

That you’ll notice is that this offset corresponds with the difference between your time zone and UTC (Coordinated Universal Time), the universal time standard. As in Power bi service the time is appplied to UTC time zone. You can refer to the bolg to find the solution.

Regards,

Frank

[v-frfei-msft]

Hi @Anonymous ,

Does that make sense? If so, kindly mark my answer as the solution to close the case please. Thanks in advance.

Regards,
Frank

[Anonymous]

@v-frfei-msftI guess I'm not sure this is actually my problem then. The dataset was last refreshed at 2pm CST and it shouldn't have been including Feb 3 data at that point, because it wasn't the next day at universal time yet, either. Maybe my dax is wrong?