In general, the first step would be to peel away any obvious non-cryptographic encoding layers. For example, the string given in your question looks like Base64 encoded data; you should start by decoding it, and examining the result. Note that the decoded data might contain non-printable bytes, so you may want to look at it in a text editor.
Once you've reach the actual cipher layer (assuming that there is one), your next step is figuring out what kind of a cipher it is. For that, you may find the answers to this earlier question useful.
In particular, you should start by figuring out the cipher alphabet. For example, if the data is all readable text (in some commonly used encoding), but the letters are scrambled, it's probably a letter-based substitution or transposition cipher; you can tell which by looking at what the most common letters are. On the other hand, if the data is full of unprintable bytes, it may be a byte-based substitution cipher such as an XOR cipher.
There are also some software tools that can automate a lot of this process. For example, the automated cryptoquip solver at quipqiup.com can make short work of most simple text-based monoalphabetic substitution ciphers. For XOR ciphers, xortool automates much of the solving process; for text-based Vigenère ciphers, PyCrack is one of the many similar solvers.
Of course, all such tools only work on ciphers of the specific type that they're designed for. If the specific cipher you've got is something (even slightly) different, you may need to get your hands dirty and write your own solver (or even solve it by hand). There are plenty of questions on this site already on various ways to do that, e.g. under the classical-cipher tag, but if you come across something that isn't already covered (but preferably something where you know what it is), feel free to ask your own.