Upload & Sell: Off
If your goal is a natural flattering appearance in artificially lit shots you need to understand how natural light models faces and simply duplicate the key (modeling) and fill (control contrast) vectors as much as possible in the room environment with the equipment you have available.
Most speedlight diffusers are designed for use on a camera mounted flash. That limits the size that is logistically practical. They bounce most of the flash output off the ceiling to mimic the downward direction of the key vector which is the characteristic of natural light because the source (the sun) is overhead. That's our perceptual baseline for "normal" and why artificial lighting doesn't look "right" (what is normally expected) when it comes from eye level from the hot shoe.
One of the ancillary effects of bouncing the most of the light off the ceiling to create the "key" vector and highlight pattern is that the light also bounces around the room creating omni-directional wrap-around "spill fill" similar to skylight outdoors. The spill fill effect along with any ambient light is what makes the core shadows the key vectors don't reach lighter in tone.
To visualize how the spill fill factor changes the core shadow tone and contributes to the perception of softness with any modifier big or small try the same set-up indoors then outdoors at night where there is no ambient or reflected spill fill.
A small SB used on a camera mounted speedlight doesn't change the direction of the key vector which creates the modeling on the face. The key vector still comes from an unnaturally low angle producing few directional modeling clue on a face. The small footprint will not overlap onto the ceiling and walls to create spill fill in the same way a scoop or cap modifier does. It will produce dimensionally flat modeling and very distinct, dark and unflattering unfilled shadows. If the flash and small SB is moved off camera on a stand it creates directional 3D shadow clues but the shadows will still be very dark and unflattering due to the lack of spill fill when exposure is set to reduce the ambient light.
On camera raising a small SB on a bracket will make it more flattering. A bracket which raises a flash a foot or more above the lens will change the direction of the key vector from eye-level to naturally downward like the bounce-based modification strategies. The small footprint will create the same distinct dark shadows but with a bracket raising the flash most shadows will be hidden from view. But if compared with a bracket and direct flash you'll see the SB does improve the overall character of the lighting much because the improvement in modeling is the result of raising the source on the bracket, not the modifier.
If used off camera a small SB needs another source to create the fill; a reflector or wall close to the subject or a second flash near the camera axis. The shadows will still have sharply defined edges but with the addition of a second fill source from near the camera they will be lighter in tone. Comparing unfilled off axis vs. filled off axis with the same modifier the photo with the lighter toned shadows and overall lower contrast will be perceived as softer. By adjusting the key:fill ratio the shadows can be made so light they are barely noticed, similar in tone to the fill present the ambient room lighting seen by eye, or darker that average room lighting to convey an sad/angry mood of the subject or a dark dangerous environment.
The brain uses many clues created by the body language and lighting to infer a person is happy or sad or that the environment they are in is normal or not. When deciding how to light a scene or portrait the mood I want to create is the first thing I consider. Then experience creating mood with light characteristics tells me whether I want a key light placement that puts light on the front of the face and eyes (so the viewer engages and dwells on the face) or put the eyes and face in shadows to convey that the subject doesn't want to make "eye contact". Viewer will find the face but will not dwell there long, moving off to explore the rest of the frame.
The key:fill ratio controls the overall contrast gradient over the subject and scene. A photo taken with flash can include all the background distractions normally seen by eye by bouncing it off the ceiling, or isolate the foreground subjects like actors in the spotlight on the stage by controlling front>back gradient with a directly aim flash or flashes and inverse-square fall-off. Experience teaches which strategy will match the goals for the shot. You need to formulate the perceptual goals for the shot, choose the most effective strategy to create the desired perception / emotional reaction in the mind of the viewer, THEN select the tools which will make it easiest to implement the strategy.
Bouncing and cap diffusers will light up a small room like an overcast day because the light is radiated in all directions from the spot on the ceiling where it hits. But in a big room or outdoors they are ineffective and inefficient.
Scoop diffusers which bounce the light forward and up create more of a front>back fall off gradient but still work effectively when there is no ceiling to bounce.
You always need to consider the direction of light that is creating the modeling, the "key" vector and whether or not it is natural and flattering. Indoor ambient light from ceiling fixtures is naturally downward but not always flattering because the angle is so steep the eye sockets get shaded. The first thing I do when entering a room is note the "key" vector of ambient light and whether it is shaded by the brow. If it is I'll use my flash on bracket to overpower the ambient and highlight the eyes because that makes the character of the lighting more flattering.
Bouncing the key vector off the ceiling can create the same shaded eye problem if the angle down gets greater that the 45° angle of the brow above the eyes. That's why the key light needs to be 30-45° relative to the eye line and front plane of the face. Adding a fill vector to the shaded sockets with a card will make them brighter and add catchlights, but they will still be darker and seem dull compared to the cheekbones below being hit by both the fill and the key vectors.
The solution to shaded eye sockets is getting the KEY light into the eyes by either lowering the key source or raising the face up into the light. I always look for higher vantage points and will use chairs or ladders to get above the subject's then have them look up at the camera to get the light past the brow into the light.
Using two flashes gives a photographer more creative options. Having the key light off camera on a stand allows it to be aimed precisely to model the 3D shape on the front faces and objects. It can also be used behind the subjects producing the same look as backlighting in the sun. in a wide shot the addition of the rim light vector creates clues about the 3D size of the space lacking in a single flash shot with front >back fall-off and evenly lights the space.
I started with two flashes in the early 70s when using brackets was the norm for wedding photography. Logistics often dictate only using one flash. Bouncing it isn't always possible due to high ceilings in churches or ballrooms. So the "belt and suspenders" approach to cover all the bases is to keep the flash on a bracket to create the flattering downward modeling when single flash must be used. That still allows the flash to be bounced or split with a diffuser when a low ceiling makes that the better strategy.
That's why the foundation of my lighting strategies is a 580ex flash on a camera-flip bracket with a DIY scoop diffuser. When I can't bounce I keep the top flap of the scoop down so most of the light goes forward knowing that the bracket will make the direction more naturally flattering, but there will be front>back falloff and a darker than normal background when I use higher shutter speeds to eliminate the color cast of the tungsten ambient sources. If more ambience is needed in the background and I can't bounce off the ceiling because it's too high I will gel my flash with 1/2 CTO orange gell to match tungsten room and stage lighting. That eliminates the color cast problem allowing me to use higher ISOs and shutter speeds that won't cause ambient blur.
When using single flash in a low ceiling room I open the top flap of the scoop. The sends most of the light up to the ceiling which does three significant things:
It makes the ceiling bounce the dominant "key" vector that models the faces instead of the flash that is coming from the back side of the scoop.
The larger "virtual softbox" created by the spot on the ceiling makes the key vector more diffuse than the scoop used directly when there isn't a ceiling available. In addition to "wrapping" the light to create gradients a larger source also makes the highlights less distinct and harsh. Fuzzier highlights also make lighting seem "softer". In a small low ceiling room it also creates the spill fill which lightens the shadows created by the key vector off the ceiling.
Moving the source of the light up to the ceiling distributes it evenly front>back from that spot and makes it fall off in intensity from ceiling and floor. So instead of seeing the same intensity of light on the entire body in a full length view the fall-off creates a vignette that makes the faces brighter than the feet. That creates an overall gradient which will attract the viewer to the faces more than the feet of a bride wearing a big white dress.
Most of the time when I use a second flash at an event like a party I park it a corner out of frame behind the action. I used the flap-up technique described above with the second flash creating rim light to define the outline of the people and create the impression of 3D space. The combination of the flash on the bracket bouncing down off the ceiling and downward off the back of the diffuser on the bracket creates nice flattering modeling and "seen by eye" contrast on the foreground.
If I can control the angle of the key light to face precisely (i.e., the face isn't moving around) I'll use portrait style lighting with the key light placed 45 degrees above the eyes and 45 degrees to the side of the nose. That combination of angles gets the key light vector past the brow into the eyes and causes the nose shadow to fall over 1/2 the nose modeling it's 3D shape perfectly. Then when I walk around to the front for a full face view or around to the shadow side for an oblique view my flash on the camera provides the fill vector for the shadows. By adjusting the ETTL A:B ratio on the flash I can "dial in" any shadow tone and implied mood I want for the lighting.
I've been shooting with flash for 40 years so I really don't need to think about the mechanics. I start with the goal for the look and mood I want (i.e., seen by eye normal or not) then adjust the lighting ratio based on knowing what ratios will trigger the desired reaction in the mind of the viewer. A:B = 1:1 for lighter than normal shadows (for kids and women), 1:2 for the same contrast I see by eye with the ambient light contrast, or 1:3 or greater for darker shadows to convey a thoughful, sad, angry mood or imply the subject is an a darker than normal scary place.
I don't disagree that bigger is better for modifiers in most situation but my choices for speedlight strategies are based on the fact I move around when taking photos with speedlights and big modifers are not logistically practical. So I've learned to make the results look as natural as possible with a combination of small scoops, bouncing and a bracket for the on-camera flash, controlling the overall "vibe" and mood with lighting ratio from a baseline of recording the same range of tone and detail I'd expect to see by eye in the same situation, adjusting from that "fit scene to sensor range" baseline when I want lighter or darker than normally seen results.
When I want better better control of the lighting character in a static situation using all the lights on stands I use the studio lights and larger modifiers I also own, not my speedlights. YMMV. I don't use softboxes on my speedlights because my scoops work similarly when the top flap is down and aimed at the subject. I also seldom use umbrellas with my speedlights on location because in most situations it's easier to bounce the light off a wall or ceiling to get the characteristics of a larger source. For example a very easy way to shoot a group with one flash is to stand on a ladder or balcony, have the people look up at the camera and bounce the flash off the ceiling directly over their heads. The photo will be filled with well lit smiling faces which are all the same size with the same exposure and lighting pattern on all of them :-)