9 rules any data analyst should try to keep:
- First and foremost, it’s not about you. Your job is to deliver something of value to someone else. If they can’t use it, won’t use it or don’t use it, what you've delivered...and what you think...doesn’t matter.
- So, before you do anything else, think hard about what you’re trying to achieve. What are your start and end points? You don’t need to know the answer before you start, but you do need to know what a good answer looks like. Is it useful? Usable? Insightful? Make sure it passes the “So what?” hurdle.
- Never assume that the person who did the step before you did it properly. You also should not assume that they can replicate it if you need them to. (This is especially important when the person in question is your boss).
- If you ask for an answer and someone gives it to you, don’t assume it’s correct. Assume it’s what they think is correct. Don’t assume they know what they’re talking about, either. (This rule applies most to the people you trust most; they're the ones who are most likely to accidentally hang you out to dry.)
- Think about your logic steps and always write them down. Writing things down is the best way to get your hand to tell your brain to think again. Personally, I never start writing code until I’ve worked through my logic stream and drawn it out. If I can draw it, I can write it.
- Always think again. Unless you’re a genius or highly intuitive, your first guess is just that and quite possibly wrong. Worst case scenario is that your first thought was ‘quite good’, meaning you don't think more about it. If you don’t ‘think more’ you’ll likely miss any deeper truth.
- Be Logical. There should never be a gap in your logic. Each step and each conclusion should lead to the next step in your story. If there is a gap, you’re leaping.
- Never leap. It’s not what a good analyst does, ever. If you postulate, you’d better know how to ‘prove/disprove’ the postulation, otherwise you’re just sounding off.
- Never sound off. There are plenty of people around us who will do that and some are actually paid to do it. It’s not what you’re valued for doing, so don’t do it—you will only worry other people when you do. Understanding the question, finding and shaping the data to answer the question and describing what the answer means, is what you’re valued for, so do that.