So I wrote some code to test this out. I ran straight rogue vs. both 1 and 3 level dip into Champion, for both bow and rapier, with and without booming blade.
I pumped out a lot of data; it will be some work to put it all into legible graphs so I haven't done that yet. But conclusions are largely as expected:
- At Tier 1 & 2 the dip is not worth it, because the increased crit doesn't balance the reduced Sneak Attack, plus the ASI delay leaves you with a lower Dex. At even rogue levels the difference is less, but it can be over 10% on the odd rogue levels.
- At Tier 3 & 4, when both builds have 20 Dex and the sneak attack dice are increasing, pretty much a toss-up, with a small advantage going one way or the other at odd/even levels.
- At levels 17-20, with Booming Blade, the dip has a pretty clear advantage. It's still about a tie on odd levels, but on even levels the dip...with all those sneak attack and booming blade dice to crit with...pulls ahead.
- Dipping for archers is never worth it. Trivantage already hits most of the time. And because you can't booming blade with archery, the crit is never as valuable as it is for melee.
Here's the data for level 20 with Booming Blade (which had the largest delta in favor of the dip)
Average damage per attack (N = 1,000) 20 rogue, vs. AC 10 to 20:
65.651, 65.392, 65.671, 65.064, 65.101, 65.364, 65.147, 64.154, 64.366, 62.338, 61.611
Same thing but rogue 20, champion 3:
69.235, 69.409, 70.624, 70.025, 69.273, 70.704, 69.253, 69.288, 67.516, 67.178, 67.437
What I find interesting is that the delta increases at higher ACs: I assume because if you do hit it's more likely to be a crit. If I went beyond AC 20 I bet we'd see that gap widen.
Anyway, I think the conclusion is pretty clear: don't dip, and if the campaign goes on so long that you think you'll reach 20, put the last 3 levels into champion.