A giant rat approaches! Roll for initiative! Time for a fight - you win or you die. Victory brings experience and treasure and the power to move on to greater threats beyond this innkeepers basement. Defeat brings death and ... er ... the end of the game I guess? No more fun? Time to pack up and go home?
Lots of dramatic and exciting situations that we want to replicate in board games are a very "all or nothing" situations. The problem is that "nothing" can lead to bad gameplay. This is particularly true in the case of the backlash that's faced by player elimination games, but also undermines some other genres of game. It's not great to have a game about stealth where the sneaking player can make a bad play and be found early - curtailing the game before it reaches its intended climax.
Fortunately game designers have been working on this problem for a good long while, so I wanted to take a little bit of time to explore solutions that "break the boolean" and find ways to make success or failure at their core tasks more granulated.
Hanabi takes a very straightforward approach to granulating failure: A single failed incident does not spell defeat. Instead players are only defeated once they have made an error three times. Some designs benefit from a simple but powerful solution over something more complex - generally those which already adopt a minimalist but very honed gameplay in the first place.
Wizard's Academy uses a modified version of this approach - which I did not invent - but I'm writing the article so I get to include one of my games That is to have a mana pool that brings defeat when exhaused, but that also has other uses. This has all of the benefits of permitting multiple failures, but also adds a layer in which players may choose to take what are essentially "voluntary failures" to accomplish other goals. This approach best suits designs that encourage players to carefully balance risk and reward.
Prophecy takes this even further in having its failure resource directly reduce capacity to progress. Battles are fought based on your strength, when you lose them your strength goes down, if it hits zero you're dead. Thus multiple failures are required, but each also increases the chance of future failure. In order to prevent this from creating run away losers the inverse is also true: You can regain the key resource, in this case through healing potions and the like. This approach best suits designs that are somewhat narrative and encourage player's fortunes to wax and wane, rather than the slow mechanical crushing of the player executing the strategy that's 5% less efficient.
Nosferatu gives an example of another sort of loss: As the vampire remains uncaught she is able to insert darkness cards into the deck that ensure each night is longer and thus there are more cards submitted to the pile. This results in less information with which to make future decisions. This approach is best suited to designs in which the principle challenge is to find or obscure information.
Game of Thrones is unusual (though far from unique) for an area control game in that defeated forces are typically not lost, instead they're forced back a territory. Essentially meaning that a defeat doesn't mean the end, but does represent a significant loss of progress towards a goal. The reason that this approach fits the game also indicates the sort of game that it's well suited to: You are not obliged to use those pieces again in pursuit of the same goal. If you choose you can ally with the person who's defeated you and throw them at a completely different opponent. Thus this approach is best suited to designs in which players can effectively change their approach on the fly.
Ninja takes a much more brutal approach, forcing you to start again if you're caught and killed. This is termed DIAS on Twenty Sided, where it stands for "Do it again stupid" and is a terrible thing in video games, wherein you need to repeat a bunch of gameplay to get back to the critical moment. However in a board game it can be excellent because humans dramatically outperform computers in their capacity to learn. Repeating your infiltration, but with recent knowledge of how your opponent goes about catching you can be a blast. This approach best suits designs that encourage a battle of wits with another player in which the nature of their approach is a deciding factor in the conflict.
Arkham Horror takes an approach that I've decided to refer to by the name of a different game, in order to maximise confusion. With the 'Jenga' approach each significant failure adds a restriction to future play - in this case as a madness or injury card. On their own the restrictions are not too troublesome and it's generally possible to play around them, but as they add up they become an imposing obstacle. Essentially meaning that your failures build up to become a wobbling tower that will eventually fall and crush you. This approach has a lot to offer designs that are looking to make their interactions "feel" weighty and to establish something akin to creeping dread.
Fury of Dracula does something interesting for a stealth game, in having an entire game within a game. In a lot of stealth games where the sneaking player is caught that's it, game over. Fury of Dracula takes a different approach: When he's found a relatively elaborate mini game is played out in which he has a fight with whichever hunter has caught him - depending on their relative levels of preparation this can go disastrously for the hunter! This sort of choice can add a lot of weight and complexity to a game, but also makes the earlier interactions more nuanced. Rather than optimising "What is the best move to find him?" a player must make choices like "Is it better to make the move that provides the most information to start the conflict or the move that provides the best preparation for a conflict?" This approach offers something to games that don't mind a little extra complexity in order to create an extra dimension for meaningful decisions.
I am certain that there are plenty of approaches to breaking the boolean that don't fall into these categories (or that some taxonomies could combine several of them into a single category) but it seems like a great place to start your thinking if feedback is indicating that some element of your game has become too "all or nothing".
Good luck and may your failures be by a small margin!