I'm interpreting priming as like random noise, since we don't know which direction it will move my estimate. Assuming my rational decision process is better than picking a result at random, adding random noise to it will on average lower its effectiveness.

If the decision's important enough to be worth some trouble, you could think about it in many different contexts, or perhaps with an intentionally varied set of primed stimuli, and seek a sort of sum over your impressions. This seems plausibly more reliable than either "avoid all outside influences" (and hope your uninfluenced decisions are best) or "expose yourself to particular random priming" (and hope that a randomly influenced decision is best or harmless). This is probably a stronger or more complex effect than "priming", but I know I often see usefully different things about an idea when I travel, or when I show it to someone else and suddenly imagine how it might look to them.