First there is the two-part structure. Part I deals with Holmes' investigation and solving of the murders of Enoch Drebber and Joseph Stangerson; Part II is a long (and, to some, tedious) flashback providing the background to the murders, followed by a conclusion which discusses the fate of the murderer and allows Holmes to explain how he unfolded the truth. Doyle believed like Gaboriau (and Poe) that the detective story is primarily a demonstration - the great detective takes on a problem that baffled everyone else, solves it as easy as pie and then explains how he did it. Works well for a short story; not so much for a novel - it needs some fleshing-out to be palatable, and turning back the clock is as good a way as any. Doyle borrowed the technique from Gaboriau and used it again - and much more successfully - in The Sign of Four and The Valley of Fear.
Doyle's "demonstrational" approach to mystery writing also means he has not much interest in the guilty party's identity. The murderer might be anybody, and turns out to be a character never seen or heard of prior to his designation as the Man Who - a device Doyle would use liberally in his later work. So secondary is the matter to Doyle that he gives it away halfway through the book - obviously, Drebber and Stangerson's acts in Utah were of greater significance to him.
Had Doyle written two decades earlier, none of this would have been of much concern. The problem is, the mystery genre had moved a great deal forward by the time Doyle wrote A Study in Scarlet, and he apparently didn't take notice. Though neither Green or Hume was a match for Doyle in terms of literary skills, they both had showed that it was possible for a novel to focus on a mystery and its unravelling in a linear (well, almost) fashion; they had also realized the naming of the culprit was a climax in itself which worked even better when said culprit turned out to be one of the members of the cast rather than a rabbit pulled out of the hat at the last minute.
Because he was a mystery writer out of necessity rather than vocation and held his work in the genre in pretty low esteem, Doyle never really cared to 'evolve' over the years. Still, he showed at times a more modernistic approach to his craft. The Hound of the Baskervilles, probably the only later Holmes story he wrote with pleasure, adopts a "modern" linear structure and for once the whodunit element really matters; it's a mystery why Doyle didn't seem to learn from this achievement and later reverted to type with the admirable yet archaic The Valley of Fear.
None of this should be taken to belittle Doyle's contribution to the genre and to literature at large. For all their occasional (and, on second thought, relatively minor) archaisms, the Sherlock Holmes stories basically set the tone for all of the detective stories to come - even hardboiled writers more or less adopted Doyle's template. Holmes is a wonderful creation and the stories bear multiple re-readings with no sign of wearing out. It's no exaggeration to call Doyle a revolutionary, one of the very few genuine ones in the history of the genre, though it's certainly a paradox.