I wouldn't put a huge amount of faith in this "poll" TBH. The methodology is quite unique (matching electorate demographics with the results of a one-off survey) and we have no way of calibrating the results with other polls in a time series.
I suspect it may be OK on traditional Lib/Lab seats, but "profiling" independent and even UAP/PHON voter demographics would be very difficult.
OTOH, the YouGov/NewsPoll polls are amongst the more reliable. 80 seats would be a great result.
