In many visual novels "choices" have very little impact on how the game plays out. These games offer a single path through the story. Sometimes they have some dead-end or side branches which all either end the game in failure or link back to the single path. Sometimes they offer multiple parallel storylines (e.g., multiple girls you can interact with), but none of these impact each other. 

For The Perils of Bianca I wanted the player's choices to actually matter, and make the overall story play out in significantly different ways depending on the choices made. I approached this as follows:

Significant choices occur mostly in Acts IV to IX. In each of these acts Bianca goes on a mission (though she may skip missions if she wants). The missions can be played in any order (except for the final mission, which is Act IX). During each mission Bianca is presented with between two and seven "situations" which she must resolve. Each of these situations has at least three, but usually many more possible resolutions. How Bianca resolves a situation tends to influence the follow-up situations within the mission, and often also affects follow-up missions.

Some of the ways in which Bianca may resolve a situation are using an item, negotiation, smart talking, threatening, performing a certain action, or sex (though sex, if possible, should always be optional). Usually it is a sequence of several such decisions, which turn the handling of the situation into a small bit of story. 

There are multiple possible outcomes of handling a situation. They may simply progress the mission. Bianca may have acquired an item. Bianca may have caused the death of a potential ally. Bianca may have antagonized a potential ally. Bianca's reputation in the game may have taken a hit. Bianca may have reduced the magic powers of the queen. Or Bianca may have gained something that pays off later in the mission or much later in the game.

I think that overall my approach to making choices that matter works reasonably well. You will notice this especially if you do multiple playthroughs of the game: if you approach missions in a different order and/or make different choices, you will see things play out in very different ways. From Act IV onward you may actually see almost no overlap between two playthroughs. I like this in a game. Moreover, it means that if players want to avoid certain content, they can. For instance, it is never necessary to engage with futa content (though I think that some of the most fun scenes in the game concern futa content).

There are, however, at least three big disadvantages of constructing a game in this manner. 

The first disadvantage is that I had to create many different ways to play through each situation. This entails not only different texts, but also different images, and different variations of many images. For instance, pretty much every image in the fairy tale missions had at least two variations: one of Bianca on her own, and one of Bianca accompanied by the appropriate heroine. Very often I had to create many more variations, up to ten of them. For some choices I had to create sequences of a half-a-dozen or more images which the player would only see for those particular choices. The consequence is not only that I had to do a lot of "extra" work, but also that in a single playthrough a player experiences a lot less than half the content of the game.

I do not mind creating scenes for a game that few players see. In fact, I like the idea that a player creates their own experience, which may not overlap with the experiences of other players. But in quite a few places in this game I personally highly prefer the consequences of one choice over the other choices. For instance, in the story of Jaclyn, the first situation you have to deal with is getting the magic beans from the farmer. The quickest way of doing that is just exchanging them for any item that Bianca possesses. A more interesting way is offering the farmer some quick relief. However, the real meat of this situation is in searching for the escaped cow, and what may follow after that. So that is what I hope that most players do. 

But if that is what I hope, why then would I create the other possibilities? Because if I only offer the players the choices I prefer, there are no choices. However, offering choices, in particular choices that matter, automatically means that some choices will be preferable over others. 

This entails that if a player wants to experience all or most of the interesting content, they have to do a lot of backtracking, or they have to do multiple playthroughs. I hope that players find the game so interesting that they want to do multiple playthroughs to see most of what the game has to offer, but I shouldn't count on that. Only those who really like the game will do multiple playthroughs (and they will be rewarded for it), but I guess that is a small chunk of the game's audience.

The second disadvantage is that it was very challenging for me to take all the complexities of the game into account. For every scene I had to keep track of the many different game states that could exist while the scene was playing out. In some cases this was not too hard. For instance, whether or not Bianca can use an item depends solely on whether or not she has that item in her possession, so either she gets the choice to use the item, or not. I just needed to make sure that there was always an option to handle a situation which would not require an item. However, in other cases this became really hard. To give an example: in Aurora's tale, six major game states may exist after the bridge has been crossed (comprising who managed to cross and what clothes they wear after crossing), which all impact what happens in this mission after that.

The third disadvantage, and probably the most pressing one, is that testing the game is quite hard. It is pretty much undoable to test each and every path through the game. For instance, in Act II Bianca has to choose between offering the dwarves to be their housemaid or to be their personal sex slave. This choice affects the conversation that Bianca has at the end of Act III, the end of every mission in Act IV to VIII, what happens if Bianca skips a mission and what she does instead, and what the dwarves will do in Act IX if Bianca asks for their help. And the choice between housemaid and sex slave is only one of the many minor choices that have a long-lasting impact in the game.

Of course, I tested each of the missions separately while developing them, but I also had to test complete playthroughs. I defined a whole set of possible playthroughs that I needed to test (e.g., do rescue mission first, do rescue mission last, do rescue mission first but still do all missions alone, skip the rescue mission completely, do only the rescue mission but not any of the other missions, let all heroines die, let one heroine die and antagonize one and leave one in prison and take one on the final mission, etcetera, etcetera). All in all I found that I had to do about forty complete playthroughs to cover about 80% of all possible variations. And if I would discover a major bug in such a playthrough, I would have to redo it (or part of it) after fixing the bug. You can imagine that that happened regularly.

So, if I create another visual novel, would I introduce the same complexity of choices that matter? 

I like the idea of choices that matter, but I would probably not go overboard with them as much as I did with The Perils of Bianca. I would severely reduce the number of minor choices that have long-lasting effects, I would have a simpler story overall, and would try to exclude options that I think are a lot less fun than others. Ultimately, that would reduce development and testing time, and would not impact the fun of the game much. In fact, potentially it could enhance the fun.