I don’t know if the Conan ruleset has anything like Star Trek’s GM guide, but there are some pertinent quotes that may be relevant (Threat is the equivalent to Doom):
A tool for pacing and dramatic storytelling aside, Threat is best thought of as an abstract representation of all the things that might go wrong at any given moment. It’s all the unknown variables, and the unfolding conspiracies, and the malign influence of powerful foes, and the clamor and clangor of battle, and the chaos that can erupt when tension runs high. It’s Murphy’s law (“anything that can go wrong, will go wrong”) quantified, and it’s the bullets in Chekov’s gun (a dramatic principle whereby elements introduced into a story must pay off or they are superfluous).
When you gain or spend Threat, the resulting effects can serve to draw attention to specific themes and ideas, or to certain elements of a scene, or to an individual character. This, in turn, creates new opportunities for the players to engage with the situation. In general, using Threat to highlight things can make the game more engaging, compared to simply using Threat to make things more difficult.
Because Threat is in large part provided by character action, it helps provide a balance to the game, ensuring that the “pressure” the player characters apply to a situation is met with equal force by their opposition. Threat serves as a visible cause-and-effect for players taking risks and facing the problems that come along with it. Your use of Threat feeds from circumstances in-game: the protagonists push their luck, and fate pushes back. This connection provides the players with an understanding of how their choices influence the problems they face and gives you a clear license to create those problems on the fly.
The causality of Threat means providing a narrative link between the ways that Threat is gained during a scene, and the effects that come from spending Threat. It doesn’t have to apply to every time you gain or spend Threat, but a few key instances can help a feel of continuity and verisimilitude.
A good example of causality might be a character adding to Threat to buy dice when climbing a crumbling cliff, and you spending some of that Threat later to make the cliff begin to subside, or enemies sounding an alarm, which adds Threat points which you can then spend to bring in reinforcements.
There’s a 10-page section about Threat in the book specifically, but hopefully the above provides enough info for you to give some thought to.
I think the last part is really pertinent to your questions. Sure, you can plan for extra bad guys to show up or for a bridge to collapse or whatever, but if a player explicitly does something that adds to threat… the threat spend is a direct result of their action. Which is especially nifty if the players do something or go somewhere that you haven’t planned for—now you have a way to respond built in to what they do.
The “spotlight” section is interesting, in that it encourages threat use to drive players to an intended solution (in certain cases); ramp up tension or draw attention to a specific part of a scene; or use threat to spotlight the abilities of a player character, by creating situations that they can solve (it explicitly notes not to create problems that are meant to stymie specific characters too often, as it will make the game feel too adversarial rather than collaborative).