Fair.
I agree it has strong potential. But how does that work out in practice?
Well, obviously it would have to be tested. But, from a gut reaction, I can't see it being too much of a problem. Giving up damage for potential damage means that the potential damage has to be greater than what you're giving up. Otherwise, there's no point, it's just a suckers bet.
And that's largely my problem with the fighter warlord builds that trade attacks for granting attacks, such as the Battlemaster currently. Sure, if you're a sword and board fighter, it makes sense to trade an attack for the great sword wielding barbarian. Fair enough. But, fighters are generally pretty decent DPS. My archer fighter with Sharpshooter is never going to grant attacks to anyone because it would virtually never be worth it.
But, since all rogues deal similar damage, unlike fighters, you can base off the rogue's sneak damage pretty easily. If I'm giving up 2-12 damage (2 sneak dice) to grant an attack, then that attack has to do more than an average of 7 damage. Easy peasy. But, because I'm giving up 7 damage on average, any bonus attack, while dealing more than 7, isn't likely dealing a whole lot more. Plus, as an added bonus, I'm still making an attack. My rogue is still directly contributing to combat, which keeps the rogue player engaged as well.
To me, it just seems like a really nice balance. A 7th level rogue sneak attacks for 2d6 damage and grants the greataxe wielding barbarian a shot to deal, say, d12+8. I'm only, really, adding about 7 points of damage for the round, which isn't going to break anything.
Granted, you might have two rogues in the party, but, then again, that party has its own weaknesses anyway - Glass canons.