The terms "NW flow" and "SW flow" are very broad and generic, so your uncertainty is appropriate. Yes, in general, when people refer to "[Direction]-flow" events as they pertain to severe weather, they are referring to the mid-level (generally 500 mb) synoptic scale wind direction.
To help your understanding, don't forget the basics. You can assume winds above 700 mb everywhere in the CONUS are geostrophic, and (generally from 925 and above on the Plains, with small exceptions). So southwest flow implies a geopotential height gradient from NW-SE (in order of increasing heights). While that doesn't necessarily imply there is a closed low nearby, it does suggest there is a height minimum/trough somewhere to your west/northwest/north (depending on the exact direction of flow and the shape of the height field, which is what you really should look at instead of the wind pattern itself). Northwest flow, on the other hand, means the geopotential height gradient is rotated 90 degrees clockwise, so that the gradient is NE-SW (again, in order of increasing height). That can indicate a height minimum somewhere to the east/northeast/north, again with details depending on the structure of the height field itself.
Oftentimes when you have SW flow and a height minimum/trough to your west, even northwest, it tends to move east or northeast, as dictated by the planetary scale westerly flow in the mid-latitudes. The reason a low is more likely to move northeast than southeast is due to the occlusion process - when that happens, the low tends to pull back towards the colder air as it becomes dislocated with the synoptic scale warm sector. Given warmer air is always south, that means it pulls north (while still getting pulled east).
When the mean synoptic flow has a substantial northerly component to it, lows can move to the southeast. That is the crux behind a NW flow event - a southeastward sinking trough with a southeastward sinking surface low and cold front.
In typical southwest synoptic scale flow, the shear for severe storms comes from the discrepancy between low level southerlies or southeasterlies that are typically forced by a surface pressure trough in the lee of the Rockies and the response to the mass divergence ahead of the upper level trough that promotes southerly low-level geostrophic flow with the southwest mid-level flow itself. In the US, the Rocky Mountains, and in particular, the contintental divide and gradual/nearly homogeneous downward terrain slope east of there artificially drives pressure falls several hundred miles east of the divide, where the best combination of warming from downsloping winds meets the edge of Gulf moisture advecting from the southeast. That is why you can get "lee troughs" without much of a synoptic scale trough above (because really all that is required is "cross-barrier" flow, which in this part of the world is westerly flow, which is generally a daily occurrence regardless of whether a trough or ridge are present, and stable lapse rates, which are also very common), and hence additional forcing along the Great Plains dryline (but this also can drive surface winds to turn from where they normally would be in a flat-land scenario).
In NW flow, on the other hand, everything is rotated 90 degrees, so you usually have southwest surface winds interacting with the northwest mid-level flow to get directional turning and shear for severe storms. As a NW flow trough approaches, you will typically see a SLP pattern with a SE-NW pressure gradient or even a S-N one, which, when combined with the turning due to surface friction, typically leads to southwest surface winds rather than southerly or southeast surface winds. However, if a closed surface low forms in a NW flow event (strong enough lift concentrated over a small area), you can get backed winds (SSE or SE) very close to the low itself. It is also possible that if the surface low picks up a surface baroclinic zone, you could end up with a SE-NW oriented warm front that approximates the future track of the surface low, and north of that front you can end up with southeasterly surface winds as well. While it can be more difficult for quality moisture to move in to a setup featuring NW flow, NW flow events typically occur during the mid-summer when a mT air mass has already been in place beforehand, so little additional moisture advection is needed for sufficient instability. Even so, if you can find an air parcel trajectory that gets back to a moisture source (like the Gulf of Mexico), you can still get moisture advection even with seemingly unwanted surface wind direction.
The shape/structure of the geopotential height field and shape of the mid-level vorticity include details that make the difference between the various cases described above. You should look for these details and examine how they impact things. For example, compact shortwave troughs tend to birth compact surface lows which can be a lot of fun to analyze for severe weather and/or chase, as they can give you excessively backed surface winds and semi-circular hodographs without destroying the thermodynamic profile! Large/broad troughs, on the other hand, tend to birth broad surface troughs that may not even have a point-minimum MSLP center, but rather an elongated trough (or sometimes, double-minima in MSLP with a trough connecting them...sometimes referred to as "banana lows"). In those cases you're more likely to wind up with a sweeping cold front no matter what else is above it. However, you also tend to wind up with a much larger area with sufficient flow and shear for severe weather, and if the warm sector is large enough, you can still end up with a severe weather outbreak if you can get storms to fire ahead of the synoptic front (e.g., from small-scale boundaries due to terrain or OFBs from prior convection, or if a "lead impulse" can set up some kind of pre-frontal surface trough).
You should look for embedded "impulses" within larger troughs. They are actually quite common, if you know how to identify them. Nearly imperceptible ripples in the synoptic scale flow can trigger a dynamical instability and develop into their own embedded surface lows which can impact the coverage and intensity of convective storms on any given day in a way that would not be present without them.
