Alot probably depends on the complexity of your 'cover' game mechanics (variations of blocking of movement/firing/seeing) and different types of weapons (ie- area effcet vs direct fire/indirect fire), and even the enemies sensor ability (cant shoot what you cant see - 'cover' as per blocking accuracy or detection). Also it can be affected by movement mechanics (can multiple enemy all move (stack) at the same spot to get a 'good' shot versus a terrain chokepoint allowing fewer individual 'units' to be able to advance and/or fire.

A classic is the canyon ambush where one side is obscured (in cover) up above and the enemy is restricted in its movement and in the open to attain its goal (and would also have to fight back uphill).

Another is the most feared 'street fighting' and interiors with blind corners - where lack of information and 'sticking' your head around a corner' can be instantly fatal.

Two different types of influence maps : predigesting the map (ie - expensive things like line of sight analysis or point to point estimated movement cost/difficulty) done ahead of time, which then serves as data to combine into immediate analysis maps (handling dynamic things like multiple enemy positions and their potential future positions). Of course you have multiple inluence maps for different analysis to be combined as per the current solution being sought.

-

There would be meta level decisions affected by things like how long it takes to assume an advantageous position (or preparations of a defensive position) which can be hampered or impossible if already subject to enemy fire (or getting to a 'good' spot does). Overall goals likewise would be considered - full-retreat / fighting-retreat / hold-to-the-last-man orders etc... adjust tactical goals.