I have noticed more difference changing the output tubes.
@ellajeanelle This has been my experience too. Especially recently, swapping from KT88 to KT120 (in amps that support this) effects a huge difference. Swapping between different brands of KT88 also makes a large difference, though not quite as much.
That said, yes small signal tubes can make a significant difference as well. Especially whatever tubes are tasked with driving multiple KT tubes.
Many folks claim small signal swaps are more cost effective since NOS / vintage types are still reasonably available, and these tubes can last several times longer than their power counterparts. And certainly, amps that use an octet (of more) or KT88 (or whatever) really rack up the tube bill! I just swapped an octet of Tung-Sol KT120 in place of Gold Lion KT88 for a VAC statement 450S. Made a HUGE difference. I also swapped its sextet of 6SN7’s (mix of NOS and new). That made a smaller (but still large) difference - and yes I tried many various combinations! In other amps I have preferred Gold Lion KT88 over KT120. And EH KT88 also sound different from GL KT88, despite being made in the same factory - just a matter or personal preference (GL are voiced more "audiophile").
The KT88 and KT120 are close enough to be reasonably compatible (bias, loading etc), but different enough to sound quite different. It’s a fun swap :)
Just clear it with your amp maker first. Anything running a KT120 near its limits would fry a KT88, and KT120 draw a bit more heater current from the power transformer.
As for JJ - I had a run of bad JJ encounters in the past (~15 years ago) and never gave them another chance since then. My rule for the longest time has been to bin JJs immediately, whenever a component arrives with them inside.