It is not my intent to get into a debate over semantics……
I often hear the term ‘Demo’ used.
Usually it is in reference to one of two contexts.
Either as a sample of music used to ‘impress’ others listening to a system, or as a means to evaluate the performance of any system.
I will infer from your post that your intent is in regard to the latter context.

I prefer to use live recordings as reference to determine if a system is performing to a satisfying level. Most are recorded at venues I attend on a regular basis. There are some studio recordings I use too, simply due to the fact that I am V familiar with the recording (many years of listening to the same version), they are well engineered, & I have auditioned them on many, many different systems.

Just to list a few….

Gene Harris – It’s the Real Soul
Ahmad Jamal – Ahmad’s Blues
Nitin Sawhney – Beyond Skin
Beethoven Piano Concerto No 5 – Anton Dikov, Piano
Billy Taylor & Gerry Mulligan – Live at MCG
Blue Rider Trio – Preach’n the Blues
Duke Ellington & Johnny Hodges – Back to Back
The King James Version - Harry James & His Orchestra
Bill Evans Trio – Waltz for Debbie
Many other classical pieces that I will not trouble you with

enjoy,
Bill

I don't even use entire songs for the initial auditioning of gear. There are certain parts of certain songs that must be difficult to reproduce because a lot of gear simply can't unravel it and that saves me the time of doing more serious listening. For those components that pass the first stage, I listen for hours on end over long periods of time with no particular music, other than I'm intimately familiar with it.

I don't try to listen for anything in particular until some aspect draws my attention and then I can match it up with music that would be more illuminating given that character.

The one deal breaker for me, and the reason I have a two stage process is that once you've heard instruments well resolved and notes particularly well rendered, you will not be happy in the long term with any component that masks those instruments or notes no matter what other aspects the component may do well.

Dynamics are at the top of my list and the ability to resolve low level detail from the mix of other things going on is job one for any component in my book. You simply can't "unhear" music you've heard.

You nailed it, Catch22.

In the opening 15 seconds of a track from the Duke Ellington & Johnny Hodges – Back to Back, for instance…….

There is a percussive passage of drumsticks played on the rim. It nearly sounds as though it is being played on a wood block. On even V good systems, it will sound like a monotone riff. Sort of like….

ratta tatt, ratta tatt, ratta tatt-tatt-tatt.

On better systems, it goes more like……

Ratta TAH Tatt, Ratta TAH Tatt, Ratta TAH TattAh-TattAh-TattAh

I realize this is a rather cryptic way to describe it, so I apologize if this makes no sense.
Nevertheless, what is changed is the subtle way small nuances are being conveyed.
There is the discovery of variation in inflection, variation in the force of behind each strike of the stick, variation in timbre, variation in rhythm, and variation in the emotion that the riff brings forth from the recording.

Bill