Well, after looking at this thread for a while, I'm not sure there's anything that I can bring to it that hasn't already been said. Go learn your SQL, learn about architecture, reporting, integration/ETL, business processes, and just data in general.
But I will go ahead and handle the P.S. request from @Max for two reasons: first I know he's been wanting someone to do if for while, and second hopefully someone will gain some value or insight from it
WARNING: WALL OF TEXT HEREAFTER
So I have no formal training in BI. Zippo, zero, zilch, divide by me and you get an error. I went to a .NET bootcamp and got taught the 3.5 test and managed to pass. I stepped out into the fast-paced world of web dev jobs with a general knowledge of a few technologies and web architecture...and quickly found out that most companies are not in a rush to take on a junior web dev. I interviewed like crazy, tried to run gamut of recruiter, HR, and internal IT...and finally got lucky when a small company decided to take a shot on me. I could not, to this day, tell you what I did right in that interview, but for whatever reason they took a risk and brought me on and started my career.
"Hey, so, we're gonna have you do some web dev, we are, but first...check this out..." - My boss, first day, paraphrased
It was an SSRS project. No one knew anything about it. Hell, I didn't know anything about it. But I googled SSRS and it had "SQL" in the title and I remembered I had one week in the 'ole bootcamp where we did SQL. And I liked it.
So I decided, wth, what's the worst that can happen? I fail miserably and am back out on the street momentarily. But I'm hourly, so I'll at least have a few bucks in my pocket. So I went at it.
It was the. worst. project. ever. The business contracted to the railroad and the report was for their SSRS server. However the data source for the report was an old DB2 mainframe located in the depths of a complex in Ft. Worth. The only way to make it work was to write a ridiculous SQL query that consisted of about seven or eight derived tables. I still have nightmares about it.
But in the end I made it work. It took 45 seconds to run and looked like a five year old came up with the presentation...but it worked.
And that's the day I learned a very important lesson in IT: if you fix it when no one else can, congratulations on your new responsibility.
So I became the reporting guy. And I tried to do a good job at it. I went to conferences and spoke with professionals and told them my issues. @JenStirrup, who is on the twitter thread of BI guru's to follow, actually helped me to figure out that trying to read the data from DB2 was always going to be a problem. If I was presenting it in SSRS, it'd be best to get it into a SQL Server instance to boost both performance and presentation.
Hello ETL...you look super hard.
And it was. DML and DDL are no joke. Architects, real Architects like my boss and @tim and @Sonny are some of the smartest people I have ever known. Certainly smarter than me...I just try and stand next to them
But I plunged in and tried my best and learned about RDBMS and Star Schema and the Enterprise Data Warehouse (hallowed be thy name, thy will be done, thy aggregations be true). The only thing we didn't cover was any unstructured/NoSQL databases. They were a smaller company and even with my larger employers I've found it's been my responsibility to seek out information on that. I also learned about the business logic that goes along with each of the architectures and how that applies across the organization.
And I crawled my way towards better.
I created a SQL Server DB modeled after the companies current Access one and then created an SSIS package to transfer the data. (SSIS is the worst and everytime I have to use it I cry a little) After much wailing and gnashing of teeth, the package worked and the DB2 data was loaded into SQL server. Overnight my reports got faster. An office buddy showed me how MS GUIs work and some tricks to even out my visuals.
And I was the answer man. I mean, THE MAN.
It's important to pause and let anyone looking or trying to get into this field know that this, this is what has given me the career I have today. This is and dumb luck. I'll give you 50/50 odds on each.
Whenever our contact/handler/paycheck at the railroad had a question...I was his boy. What's the latest with this? How does this thing work? What does this do? What's new and exciting? It was (and I believe still is) a job requirement that I always be constantly revising and updating my skills. Wherever you come down in this, you'll need technical skills and constant learning to keep up. Even if you're a BA. (If you want to be a good one anyway)
And that means data. Learn about data. How we store it, categorize it, interpret it, see it.
Over the 5 years of my career I've been with the first outfit, some freelance work, and the current mega corp. I'm still very much learning myself. But I know it's my job to present data to my users. And presenting data means not only knowing about the technology, but also how the human mind works and interprets the world. So many of my users look for "their" data and not "the" data. I believe it's important that I know that so I can try to help them.
This is a long read (as I knew it would be) so if you read it all, thanks. If you can find some value or lesson in it, that's awesome as well. I'm a developer so I came at it from that angle obviously, but I think that the biggest lessons I learned carry across the whole spectrum of BI.
It's a great place to be and great time to be in it. And all of this is just my .02 so YMMV but I feel very lucky to have the career that I have. It's cool to be able to use that word.