DNSKEY metric at the root

Part of KSK rollover, Paull Hoffman reports change in pattern of DNSKEY traffic at the root. SHould it be tracked in a specific metric? Should we have a metric about mix of traffic at the root, per record type?