math

To be able to do good analysis, not just interpretation of pretty graphs, you need to know math.

One of the biggest variations between different companies is the degree to which math is actually used. Start-ups in particular are notorious for deciding "directionally accurate", despite often having enough data that they can do meaningful regressions and generally use statistics. Meanwhile, older companies often lack large enough sets of labeled, useful data, but still want to see how many sigmas of difference an intiative makes.

No matter the environment, learn the math. It's helpful.