ERA5 2m temperature calculation

I'm having trouble reproducing the results being published in media and the ECMWF Twitter feed about the global average daily temperature: https://twitter.com/CopernicusECMWF/status/1676934454877495296/photo/1

Perhaps I am doing something wrong?

1. I download a reanalysis of the 2m temperature from here for the last available day (July 3rd 2023): https://cds.climate.copernicus.eu/cdsapp#!/dataset/reanalysis-era5-single-levels

2. I average it out with python and print out each hourly average in Celsius 

   input = NetCDFFile("data.nc")
   data = input.variables["t2m"]
   for hour in data:
       print(np.ma.average(hour) - 273.15) # convert K to C

3. I expect that the averages would come close to the 16-17C that is being mentioned, yet my averages hover around 7-8C. I even plotted it with Panoply and am getting the same results

Am I using the wrong variable, or something else is a miss?