The Heart of BI / OLAP is your data

There are plenty of vendors eager with a sales pitch for BI/OLAP projects, eager also to give you the impression that all you need to do is buy their product. This is wrong, perhaps dangerously so, because **the heart of BI / OLAP is your data** and the core challenge is to transform your data into a form where it can be easily **and correctly** analyzed.

The principles of operation are similar regardless of which products you choose. Your toolset will consist of, at minimum:

  • An OLAP tool, with or without a RDBMS behind it
  • An ETL tool, which might be a software product or might be a set of scripts
  • Hardware to run it on, chosen and configured to server an analytic load well;
    this could be hardware you own, or a SaaS or cloud offering
  • Configuration thereof

But this analysis understates perhaps the most important part. ETL needs extensive configuration (for moderate cases and powerful ETL software, and some luck) and more likely, carefully crafted software to transform business data from whatever form it lies in, to a suitable shape for analysis.

Of course I cannot help but mention that Oasis Digital works on such projects; but regardless of us, effective OLAP involves an astonishing amount of “getting your hands dirty” digging in and understanding the precise meaning of bits of data flowing out of one or more (perhaps many more) operational systems. This work is arguably so unpleasant that it leads many organizations to skip OLAP, but that actually misses the point. The work can’t be avoided, without also avoiding the full value that could otherwise be obtained from a correctly populated analytical data store.

Alternatively, the work could be skipped, if you don’t mind incorrect analytical data and incorrect conclusions drawn from it. This doesn’t seem like a strategy for success.