An awful lot of postelection analysis on VA is using the standard consortium exit poll, as is typically the case. This may not be wise; the exit polls have long had serious problems and it is not clear that they have been corrected. After the 2017 VA election, Catalist did a stern demolition job on that year's exit poll that suggested rather strongly that both voter shares by demographic and preferences by demographic were quite far off (link below).
The Politico article points out some of the differences between this year's VA exits and the AP-NORC VoteCast survey. I am on record as having a lot more faith in AP-NORC than in the exits...and I continue to feel that way.