The least tactical conflict possible is one that presents a single goal and affords essentially only one route to reach that goal. For instance, imagine that the big bad the townsfolk have hired the party to destroy is an everyday inanimate wooden crate. To keep things simple, it's the target of an ongoing magical ritual that has given it immunity to all forms of magical damage; to have some kind of exigency, we'll say that the ritual will extend the immunity to all damage in another minute or so. Sure, there's choice—for example, the characters can decide to punch it bare-handed or to hit it with their weapons—but the tactical assessment is too easy; the characters can surely guess which would be more efficient, and they're just going to repeat that choice until the conflict's over.
There are several ways to can build such a conflict up to make it tactically interesting. This is general advice, not specific to Dungeon World, but in DW you can use the concepts you mentioned both as genre-appropriate inspiration for goals and consequences and as a way to bookkeep for these techniques.
Set Multiple Goals
We can add a modicum of complexity by bringing multiple goals into the picture. The dastardly crate has captured some children and is holding them hostage. Consequently, the party needs to both wail on the crate and untie the kids. The party obviously wants to avoid injury both to themselves and the children. Thus, they need to unload damage quickly, free the children, and try not to get any splinters. The multiple goals have made it a little harder to judge the best of several alternatives because progress towards one goal is (usually) incomparable with progress towards another.
Once there are multiple goals, it's easier for actions to have multiple, bundled (i.e., inseparable) consequences. For instance, the party should expect that diverting some members to free the kids will, though reducing their immediate damage output, mean getting some allies for the beatdown—if the kids can be untied fast enough and persuaded to help. And maybe it turns out that the party's blows don't just hurt the baddie, but that every hit launches a flurry of crate fragments, damaging the people nearby. Bundled consequences have made it harder to judge the best of several alternatives because players now have to compare tradeoffs, not just different kinds of steps forward.
Make Choices Interact
Consequences might also be indirect, putting new options on the table while permanently discarding others, and they can bring new goals to light or shift goals around. Let's say that the wizard has a teleporting scroll that can either position the fighter to attack from within the crate (where she can do more damage, but risks friendly hits) or evacuate the children (getting them out of splinter range), but not both. Whether the wizard uses the scroll or not, later choices are affected by his decision. As for shifting goals, suppose the crate is actually a good friend of the children, and they start sobbing and (if untied) beating up the heroes when they see it being hurt. Does the party go all out, hold back until the kids have been ushered away, or let the crate survive, hoping that they can rehabilitate it later? These interactions between choices have made it harder to judge the best of several alternatives because players have to consider and weigh future possibilities as well as immediate effects.
Set up Combinatorial Explosions
Finally, a lot of choices are parameterized. For example, when the party decides to hit with a sword, they also decide who's on the swinging end and who's on the receiving end. The same is true of untying, teleporting, etc. So we can add tactical complexity relatively easily by making those subchoices nonobvious. Now the nefarious crate is holed up in a warehouse along with several identical-looking henchcrates: one of the evildoers is sitting in the middle of the floor, another is perched on a platform, yet another swings from a pulley, while three more form an imposing stack against the far wall. They all share the immunity to magic, but only the ringleader is becoming invincible; killing the others can wait. Likewise, hostage children of various toughnesses and loyalties are scattered about the building. With a dash more of choice interaction—there's only one ladder—we now have a situation where players will have to think tactically. These combinatorial explosions have made it harder to judge the best of several alternatives because the routes through the conflict are too many to consider one by one.