What happens to tokens that are exiled by a flicker effect — do they return to the battlefield?

When a token is exiled, it is moved to the exile zone, but it immediately ceases to exist there. This is because tokens only exist on the battlefield (CR 110.5g). As soon as a token leaves the battlefield for any zone — including exile — it ceases to exist as a state-based action check (CR 704.5d). By the time the 'return' part of a flicker effect would try to bring it back, the token is simply gone; there is nothing to return.

A flicker effect typically says something like 'exile [permanent], then return it to the battlefield.' The 'return' part of this instruction looks for the same object that was exiled. Because the token ceased to exist in exile, there is no object to return, and nothing enters the battlefield (CR 400.7 governs object identity across zones, and a token that ceased to exist has no persistent identity to track).

It is worth noting that even if a replacement effect tried to change where the token goes instead of exile, this would not help — once the token is in exile, it is already gone. The rules treat the token's non-existence in exile as absolute.

Concrete example: You control a 1/1 Soldier token and cast Cloudshift, targeting it. The token is exiled, immediately ceases to exist (CR 704.5d), and when Cloudshift's effect tries to return it at end of combat, there is nothing to return. No creature enters the battlefield.

This also means flicker strategies do not work on tokens — any 'enters the battlefield' abilities the token would have are irrelevant, since the token never returns. If you want to reuse ETB effects on tokens, you need other means such as blinking the card that creates them.