Personally I never considered branching as "replayability". A lot of devs claim replayability when the large majority of the 2nd playthrough consists of SKIP SKIP SKIP.
Well, my idea of branching is about giving the players something else to read instead of "skip skip skip". That's why I always need more of them: ideally every dialogue should always have three distinct routes to be explored, even more ideally, picking one or the other should give something making it worth (a stat point, a plot coupon an exclusive image, etc).
Replayability is when you can pick up a game and enjoy the game as a whole again. For instance when I cleared Witcher 1 and 2 for the 3rd time.
Personally I really dislike branches that lock out love interests. The worst thing is when you get a shitty ending and have to try and figure out where you went wrong so you can reload without having to replay the whole game.
I think we are thinking two different things when we speak of branching: what I'm trying to achieve in my games is giving the chance to personalize every gameplay, and I'm doing it by giving the players many dialogue options that will subtly (or not so subtly) influence the NPCs and then making the story feel more dynamic. My source of inspiration being the great cRPGs of old like Fallout 2 and Planescape Torment but of course in a more limited scope.
The branches like "if you leave the wallet on the walkway you'll end up with the redheaded tsundere while if you take it and pocket it you'll end up with the prim student president" were nice and all 20-30 years ago, but had the tendency of having the drawbacks you mentioned, namely a gameplay mostly consisting of an "oh, a choice: let's save here, go on, have a bad ending, reload, skip skip skip..." bruteforcing with a little of guesswork thrown in.
Also, I don't plan to put bad endings to routes (I would like to add a bad ending to this game, if the MC manages to botch his investigations and doesn't solve the mystery, but it will be pretty difficult to get) and to make very clear that you are on a route or the other, with most LI-specific content happening in the game proper instead of just in the epilogue.
Unfortunately how I like to set up stories, playing a type rather than another
will lock out love interests, just as it would happen in real life. A girl likes bold, daring swashbuckling heroes like the one you're roleplaying just now... but she's the blonde innocent noble virgin princess, so boring, while you are aiming at the sultry dark lady that, unfortunately, would favour a gloomy, introspective, cynical Byronic hero that you have no wish to play.
This can be sorted out by having the lock being a soft one: it's not a single choice that blocks, due to some unforeseeable butterfly effect, a route and its ending, but the sum of many factors, both roleplay-wise and action-wise (the sultry dark lady will be swayed by a bold and daring swashbuckler if he has
some gloominess and introspection going on
and he manoeuvres her political enemies to fight each other leaving open her way to the lead of her House).
Also, if I do my job correctly (and I admit freely that I'm a
long way from there just now), the players should be able to notice it ("No, Mr Mareli, I don't think I can trust a prankster like you") early enough to be able to amend their ways to get a "Well, Roberto, I see that I can rely on your discretion... But realize that what I'm going to say is not easy for me." hopefully even without the need to reloading.
Now, the difficult part is striking a bargain with the player: I give the possibility that their favoured love interest could require making choices that they might not want to do, but in exchange I must swear to be as clear as I can be about it, so they'll know beforehand what they're getting into.
For this game, I'm thinking of expanding the uncannily specific "personality test" to give clues about that kind of things, instead of just tracking relationship points ("It seems that you go well with someone you recently met, but your kindness of heart is going to be judged negatively.")
And then there are the players who do not want to make a different choice for a branch. Maybe it was a moral choice or a choice for an unattractive love interest. Either way that makes that branch invalid as replayable content for that player.
I remember a game that I liked a lot "The book of the bondmaidens" (or a name like that) where the girl that gave me the hots was the evil bitch and required moral choices that I couldn't bring myself to make. Despite me liking the game and wanting to try different routes, I've yet to replay it, so yes, I get what you're saying.
I would prefer less time put into branches and more time put into ripple effects with fewer branches. Ripple effects are when choices you make noticeably affect events and characters. Good ripple effects make the world seem more alive.
I think about what you're saying as "having little choices but not being lazy about it and making them count". Like the awesome "Imperium Bureaucracy Hero" game that emerged in the release forum a few weeks ago.
But I prefer a less "tactical" and more "strategic" approach: namely of piling up lots and lots of minor choices, to be taken without much planning but just according to what players will feel is more fun, until the story will end up noticing.
The way I'm trying to do it is by having stats marking how you're playing your MC. Now, Roberto is pretty fixed as MCs go, that's one of the reasons I chose to let him keep a fixed name despite players hating it, but you can steer him toward one direction or the other. Do would want him to be kind or brash? Turn his witticism up to eleven or have him be discreet? The other characters will react to this and give clues or help or hindrance that ideally will end up changing noticeably the story (or just open up their legs).
That said, I love having choices. It gives the game flavour and it also gives the story a sense of being personalized.
I remember having a similar discussion in the past: I was asked if what I was doing was aimed at replayability or personalization. I say replayability because I believe that having the chance of replaying the game with a different MC's personality would be good for inviting to new gameplays, but if players will like it only for personalization alone I won't complain and will think my long hours at the keyboard (and long sleepless nights trying to get ideas) will have been well spent