Why is my Future Annual Consumption estimate so persistently high?

 I’ve now (courtesy of @ChristopherS_OVO - thanks!) had some actual figures for the calculation of my FAC, and they’re as clear as an early-morning midstream sample. I learnt a few things, though:

• The EAC is indeed allegedly updated to reflect data shared by the supplier. If that sharing fails for some reason - and it clearly did in my case - the EAC reverts to a default value, subsequently adjusted for whatever data have been successfully shared since the fault. 
• I can’t for the life of me work out where these default values come from. I’m told they are for a ‘typical E7 user’. So no attempt at all to take into account the factors that Uswitch and comparethemarket say are used to match the EAC to the user’s household [“central industry sources including the Retail Energy Code Company … estimate your consumption over the next 12 months based on a number of factors about your property and based on your property’s historical consumption.Your usage comparison is based on your energy usage compared to other households in your postcode region. Your usage will be affected by a number of factors including the size of your home and the number of people who live there.”]
• Meter readings are thus used to adjust the EAC from its initial default value, but this can only happen of course if OVO successfully submit readings. This seems to be where the fault is in my case.
• OVO then generates an FAC from the current reading and the current EAC, the proportions of each depending on the time elapsed since Week 0. By Week 52, the EAC proportion will be 0 and the meter advance 100%. 
• It’s impossible for me to work out how the EAC is adjusted when a reading is submitted, but I suspect it’s a similar arrangement.* This being the case, the FAC should approach parity with the meter advance before the year is up. How long before is a mystery, but it’s not likely to be nine months before.
• I can see no reason why OVO can’t generate a more accurate FAC if it’s clear that the EAC is wrong. OK, this would involve manual intervention, which they understandably try to avoid. However, I’m sure it would have taken much less effort on their part just to do this rather than involve at least five different employees in trying (and failing) to explain why it went wrong for me in the first place.
   

[I just realized after reading this through, though, that this was the sort of approach Fujitsu took when faults appeared in bits of their Post Office Horizon system. I won’t draw any more parallels ...] 
    

*   If it were a similar arrangement (the EAC value changing pro rata with the meter advance over a fixed period), then we’d end up with a second-order function which would give a parabolic curve if the actual consumption was lower than expected. Something like this (I know how you all love my pretty charts 🙂):