After 20+ years of technical, economics, and business analysis, I will offer up my best piece of advice to young data analysis folks looking to make an name for themselves: Focus on the mix. Or more particularly, changes in the mix.
The mix of what? It could be most anything. Here is a recent example from economics, arguing that the slowdown in wages is in part due to a mix shift in the country's employment to lower-paying industries. In the corporate world, it could be the mix of markets, or customers, or regions. Typical metrics used in the business world almost always miss mix. In a large aerospace company, we had the irritating situation that the profitability of every single product line was rising at the same time revenues were rising but overall profitability was falling. The problem was the mix. The most profitable product lines were all on older aircraft, ironically because they were the least reliable (aerospace parts companies have traditionally operated on a razor and blades model, so that more failures let one sell more really profitable aftermarket parts). As airlines modernized, our mix shifted to new product lines which were less profitable. This difference in profitability was also due to a mix shift -- since newer products were way more reliable, more newer aircraft in the fleets shifted the mix from aftermarket sales (which were astoundingly profitable) to OEM sales (which were often made at a loss).