Anyway, to the matter at hand: the Fronius does sound like it's keeping a record of what's being collected; for now we will believe what it says.  The other meter, installed by the power company, is telling you what?  Does it directly read 540 KWh or does it read something else?  You did some calculations that didn't go anywhere.
If the meter displays kiloWatt-hours, then the voltage doesn't matter much to its reading.

 642.6 less 1% wiring losses = 636.2 KWh

 636.174 less 8% transformer losses = 585.3 KWH

Without accurate on-site utility acceptance readouts the above figure gets within 7% of your estimated 540 KWh. Unless you sit and watch that readout until it turns over to next unit - or have access to utility accounting summary that is not shabby so far...

"Anyway, to the matter at hand: the Fronius does sound like it's keeping a record of what's being collected; for now we will believe what it says.  The other meter, installed by the power company, is telling you what?  Does it directly read 540 KWh or does it read something else? "

The meter is reading 000004
the multiplier on the meter is 120
The meter has read 4 for about a  12 days, and I estimated roughly that it could roll over to 5 in the next few days(8 kwh a day)

If the meter displays kiloWatt-hours, then the voltage doesn't matter much to its reading.

The meter reads in Kilowatt-hours. I just wanted to let you know that the system was commercial working at 600v and the use of transformers were needed, which do have some loss.

You're concerned about a fairly large loss; 174/714 = 25%  
I thought 25% what pretty large considering the technical data I've read usually estimates 5% for residential(of course perfect lab conditions)
http://www.solarelectricsupply.com/pdf/Fronius/ig_2000_3000_2500lv.pdf
This is the link I just found. At the shop I'm sure they have a lot more info.
We have the 3000 model installed and its output is 240 v

I'm pretty sure it controls the grid connection and it goes into standby at night
I will try to find out some more info tomorrow .

Well, in a perfect world anyways....  :)

Both nominally record output with a precision of 0.1kWh or better.

We're at around the 700 mark since install and the two readings differ from one another by several percent IIRC.  And indeed, the 'total generation' meter by whose reading I will be paid is the more generous of the two!

It's maybe a somewhat less exact science than we'd like.

Rgds

Then if/when they don't match with mine reading higher by more than a miniscule delta I'd call for repairs to their equipment and a refund of the shortage based on the ratio of the readings.  Then sue them for the shortfall (plus the cost of the meter and court costs) if they don't fix up and cough up.

Don't know how it works there but over here we have a "small claims" court for small value money-only suits, where lawyers are disallowed and the rules of evidence are greatly relaxed.  (The accused can transfer to a full-blown court if they want but it's not usually done.)