I've given this a rethink and I think using the combination mechanic could still be really fun, but based on feedback and exploring the monk in the playtest further I think maybe flurry is OK. Maybe it's not really as exciting as I'd like, but it's fine and safe and does its job.
I still like the combination strike mechanic and I'd like to flesh it out - not necessarily for the monk, though I do like the image of a monk doing combos. After some number crunching, I think this will be better:
Before the attack roll is made, the player has to announce the combo and how many hits are in it. The player then spends a number of expertise dice equal to the number of hits in the combo -1 (1 for a 2 hit combo, 3 for a 4 hit combo).
The player then rolls however many attack rolls as they have in the combo, noting which die is the finishing move's attack roll.
If all the dice hit, the combo is successful and they deal an unmodified 1d4 for each successful hit that led up to the finish move (1d4 for a 2 hit combo, 3d4 for a 4 hit combo) and a number of expertise dice equal to the total number of attacks made (2 for a 2 hit combo, 4 for a 4 hit combo) + your primary damage stat in damage.
Critical hits are still critical hits and apply to the individual attacks that crit normally.
So a 4 hit combo will have a 41% chance of success vs AC 15 at lvl 10 (based on an 80% chance for each individual die to hit or crit [75% hit + 5% crit], so .8^4). And do an average of 32.925 damage including crits. A 2 hit combo has a 64% chance of success and would do an average 17.425 damage including crits.
To be honest, that feels too low but it does have a pretty high chance of succeeding.