Because looking at 2m or 850 average temps isn't useful for seasonal models. They are useful for 500mb heights or MSLP. You can then guess what kind of weather can come from that. Also, for ECM, we see it for a monthly average only and you therefore just get to see either the strongest signal for that time period or an averaged into nothingness (which is useful as well). Generally what will happen is (if the forecast is in the ballpark), the warmer areas will be even warmer, the neutral areas will be actually cold and any areas showing cold will be even colder. You get to see this on the weeklies playing out quite often.
So ECM showing an anomaly between Greenland/Iceland for November, similar for December and then an anomaly over Scandinavia for January allows people to understand what is possible from that. However if you look at surface temperature prediction it looks rather lacklustre, but anyone knows that if things setup that way this forum will be having a meltdown.