Furthermore, if you're going to compare one individual poll with another individual poll, both have to be conducted by the same firm. If they're not, you're comparing apples with oranges, because each firm has its own 'house effects'. The 54% recorded in June 2016 was in a Survation poll, not a Panelbase poll. But as it happens there was a Panelbase poll conducted at around the same time, and it had Yes on 52%. So, whichever way you cut it, there's more support for independence now than there was back then.
The new poll is also unusual in that it shows a Yes lead that is big enough not to be considered what the Americans call a 'statistical tie'. In other words, even if the poll's margin of error is taken into account, Yes would still be ahead. As far as I can see, this is only the third time that has happened in any poll from any firm since the indyref in 2014.
Furthermore, this is only the second poll since the indyref that has shown Yes on 50% or higher even before Don't Knows are excluded. (The other one was the Ipsos-Mori poll from 2015 that I mentioned last night.) I'm not sure how important that is, because there's no good reason why undecideds should be left in. But it does give us a useful response to unionist commentators who like to portray Don't Knows as "presumed No voters".