What makes "quality" analog AV cables better than cheap cables?

IMHO the integrity of the shielding is what matters. Without shields, RFI and EMI and black-brick switching-requlator spikes are aggressors into and onto the audio.

Foil may be best, but foil is fragile.

Thus multi-layer woven copper-braid [many small wires in the braid, dense enough to block the passage of light] shields should be your goal.

Been working with a guy on piezo-sensors, located a couple meters from the signal conditioning analog-circuitry. He uses 16KHz LPF into buffers, and then 5KHz LPF into the ADC sampling Vin, with that final large capacitor providing the ADC sampling surges.

His usual noise floor, using a 60Hz nulling algorithm to reduce/remove power-line electric fields, is about 1/2 a quanta of 4,096 levels with 0/3.3 volt ADC range.

Thus the RMS is 3.3/8192 or about 400 microVolts RMS, with 2 meters of shielded cable between sensors and analog circuitry.

The "laboratory environment" includes numerous black-brick power supplies, and some unused old computers with metal cases that show 180 degree phase-shift of the 60Hz; a DVM shows 200+ volts between these unused computers and the analog circuitry/sensor returns.

Forgot --- here are several 2Ghz and 5GHz Internet Routers, about 6 feet away. I've carefully encouraged use of passive RC LPFs (15KHz and 5KHz) to reduce high frequency energy into the opamp buffer which might overload input-diffpair isolation-diode junctions; for these LPF's to be effective, the impedance of the LPF's shunt element, the capacitor, has to be low inductance with the capacitor's RTN node tightly wired to input-signal's RTN as well as the output (filtered) signal's RTN).

Thus a comprehensive approach to protecting your audio signal cleanliness is your task:

1) cabling, with densely-woven multiple-layer copper braid shields

2) cable connectors to/from the circuitry, with very short connector--PCB transitions to minimize that loop-area (magnetic field pickup)

3) filtering, to exclude RFI

4) proper use (skilled use) of mechanical structures (planes and cases) to ensure the interference must attempt to pass thru filters, and not be easily routed around the filters

I recall a discussion about a transistor preamp, where a capacitive-path exited from the OUTPUT pin all the way back to the INPUT transistor; the product reviewer showed (by testing) the injection of AM_RF energy into the OUTPUT pin would cause intermodulation (rectification) in the input transistor. Other preamp designs (this was RIAA phono preamps) did not have that capacitive topology, and those other preamps had BETTER reputations in the marketplace. An output RFI filter would have made the difference.