It isn't.
Or, it is and it isn't. The beginning of Linda Hutcheon's textbook 'A Theory of Adaptation' discusses the key concept of fidelity in adaptations, i.e. the resemblance and 'faithfulness' to the original work at hand. Much of the discourse around translation theory pertains to what degree a work resembles its predecessor, and to what degree this is a desirable trait. Both contemporary theorists and historical figures (many on record since Plato started discussing mimesis of his 'forms') have predominantly tended towards a 'literalist' position of adaptation -- placing "axiomatic primacy" on the source -- despite the fact that the most literal form of a work is the original. To have a completely literal adaptation would necessarily be a copy of the original, as any translation or adaptation by its nature must go through a transformative process, a process of appropriation and interpretation, in which the story is filtered through the redactor's "sensibility, interests, and talents." The novelty of a story thus lies in what exactly is done with the original text, "repetition without replication," even though theorists may acknowledge this and subsequently ignore it, demanding a very weighted balance between fidelity and novelty, when novelty is mentioned at all.
However, as Hutcheon quotes Bluestone, when an adaptation is a financial success, we tend not to question its fidelity. The capitalist approach almost always trumps the theoretical. I find it interesting that adaptations are seen as the safer bet in popular culture, even when they are derided for lacking originality -- perhaps this pertains to the idea of pandering to the lowest common denominator, and this conflict with 'high-brow' versus 'low-brow' culture. It would be quite interesting to see a statistical study on what fraction of the fiction bestsellers market is made up by adaptations (anecdotally, I have seen far more successful adaptations in the 'Young Adult' market, especially for teen girls, as opposed to the 'Adult' market, which also perhaps feeds into the idea of adaptations being juvenile -- the abhorrence against anything associated with teen culture is a topic covered much more in depth by other scholars). If one stretches the definition of what an 'adaptation' is (see Fischlin and Fortier, who view it as any alteration of cultural works, which lends to an interesting discussion on the difference between 'adaptation' and 'an adaptation'), could we perhaps see literary 'trends' as adaptations of a sort (e.g. the plethora of 'Gone Girl' or 'Hunger Games' imitators, and similar), taking the essential characteristics of a successful work and transposing them?
Harkening back to my interest in Welsh mythology and the bardic tradition of Celtic cultures, I see interesting comparisons between the medieval poetic/prose corpus and today's storytelling 'worlds.' The more bards and storytellers of the period shared and adapted work, often simultaneously, the greater the dissemination and recognition of the tales would be, enabling listeners to identify and love the stories, gaining pleasure from "repetition with variation." Listeners could pick out the nuances and differences inherent in a storyteller’s technique that would otherwise go unnoticed with a single listen of a single story -- this is a benefit, as few people would thus be insisting on fidelity to an ‘original’ if many different forms are consumed, granting greater flexibility and leniency in error to the bards' oral delivery.
A contemporary example I am playing with the applicability of is that of the 'Marvel Cinematic Universe' ('MCU'), which adapt from many difference comic book sources, all of which overlap and borrow from one another and have had different iterations under different authors. This is a gateway for fans of the MCU to get into the back catalogue of Marvel Comics, where they can then understand the multitudinous nature of the 'originals,' allowing for appreciation of the new ways that the stories are told over the years. The counterargument to this comparison is that comic book fans are notoriously finicky (or, the most vocal critics have notoriety, while the majority of fans who are quite fine with adaptations are not so frequently heard) about the fidelity to their chosen version of the story being told. Look at the online fervour that springs up when a beloved character has their gender or race changed for the tastes of a new audience (though, when such a change is 'successful' -- e.g. the white character Nick Fury being played by the charismatic Black actor Samuel L. Jackson -- we again hear fewer complaints, as previously mentioned). Speculatively, this could also have been a similar case in medieval Wales. Listeners of the era, either through limited exposure to multiple versions of tales or due to human bias favouring the first version encountered, might settle on one version and be disturbed by other versions. Think of the frustration of cognitive dissonance when (for example) a retold version of a Greek myth does not match the one you heard first: "That's not how it happened! That's not how his name is spelled! Those characters never lived at the same time!" you might cry. Therein lies the danger of adaptation: personal taste. Next I'll be moving on to what things are changed in an adaptation and why, leaving aside concerns of fidelity to fans and theorists.
So. Conclusion. Fidelity isn't important. Except when a person decides that it is, whatever their reasons.