Sam mentioned that we had talked about this in the past. I certainly find that there's a coherence to the 'top down' approach that it's hard to get from a more individual approach. For a start...typically the mix will need some highs and high mids. It's not a totally hard and fast rule of course...someone might have added 6dB of top to every single source at tracking.
But I can't remember the last mix I didn't add high end to. Obviously you want to add the right amount, and you can't really judge that by adding it to individual sources...you need to hear how the whole balance responds as it brightens. You'll soon find out if that works, or if it just needs to be brighter drums, or guitars.
Anyway, FWIW, here's a breakdown of how I personally approach this, from an interview I did with Acustica...
"I start with EQ, and I will typically see what needs to be added in the way of top and bottom end. Broadly speaking, I’m looking for a high shelf at around 10 or 12 kHz, a broad bell at around 2-3 kHz, often from White2, and a broad bell at around 30-60 Hz depending on the key. I may nudge some sources up or down as I begin to shape the whole spectral balance of the track. I’ll apply some bus compression, often Pink2, again, and a pre module at the end of the chain. I’m not looking for dynamic control really, but more the sound of the compression, the envelope. To some extent the pre module will affect that too, and it may also prompt an EQ tweak. So, I cycle around those plug ins, nudging the balance at the faders if needs be, and getting the shape...I’m looking for a treatment that suits the track as a whole, that flatters the majority of the content, and that’s appropriate for the genre. Some elements, like drums, may well have their own bus treatment - they might need a push that’s too much for the mix bus. I want to keep hearing the track as a whole.
Once I have that balance and spectral ‘shape’, I will then go back and start making corrections to the sources, applying effects. I might take a little top or bottom off a source that the bus chain has over-enhanced in those frequencies. A good example might be an acoustic guitar, or vocal that has become a little bright. I’ll generally automate a lot. As the process goes on, I’ll start looking to create points of interest in the mix, effects throws, rides, cuts or edits etc. I’ll push the mix bus fader to create impact on the choruses. I may go back to the bus chain at some point, and make a tiny change - a squeeze more top or high mid.
I tend to EQ sources pre compression, but as mentioned above, there’s always more EQ downstream at the bus stage. So it’s often both. I don’t want the source to hit the compressor in an unbalanced way, so there might be some correction for lumpy frequencies initially, but then a lot of the broader shape comes from the bus chains, while the majority of the dynamic control will be at the source track. All that said, in terms of the final sound of the mix, the EQ is before the compression"