If you run a business, you've probably been told that you need to be "data-driven." Fair enough, but nobody explains what that involves on a Tuesday afternoon. Here's a short, jargon-free version.
What analytics is for
Data analytics means using the information your business already produces to answer questions and make better decisions. That's all. The information might be sales records, website visits, customer messages, delivery times or ad spend. The questions might be "why did sales dip?", "who is likely to buy again?" or "which of these two offers should we run?"
The four kinds of analysis
People talk about four levels. They build on each other, and most businesses need the first two before the others.
What happened? Descriptive analysis. Sales by month, orders by product, visitors by source. Most reporting lives here. Valuable, but it only describes.
Why did it happen? Diagnostic analysis. Sales fell in March. Was it one product, one region, one channel? Did prices change? Here you dig into the causes.
What will probably happen? Predictive analysis. Forecasting demand, spotting customers who are likely to leave, estimating how many orders to expect at festival time. It uses statistics or machine learning on past patterns.
What should we do? Prescriptive analysis. Recommending an action, such as how to split a budget or what price to set. It's the hardest and usually comes last.
If your reports still disagree with each other, don't skip to prediction. Fix the first level first.
What you need to start
Less than people think.
A few clear questions. Which products make money? Where do customers come from? Who comes back? These guide everything else.
Reasonably reliable data. It doesn't have to be perfect. It has to be consistent and well enough understood that you know its limits.
One person responsible. It could be an analyst, or a manager who likes numbers and has time. Someone has to own it.
A routine. A weekly look at a few numbers, with the people who can act on them.
You do not need a data lake, machine learning or a six-figure software contract. Those might come later.
Where the time goes
If you ask analysts, they'll tell you that most of their effort is on the data, not on clever analysis. Pulling it together, cleaning it, working out why two systems disagree. That's normal. It means the best investment for many businesses is boring: get your tracking right, keep a clean table of orders, and agree what your key metrics mean.
Common mistakes
Collecting before deciding. Gathering everything "just in case," then never looking at it.
Too many metrics. If everything is a key number, nothing is.
Confusing correlation with cause. Ice cream sales and sunburn both rise in summer. Ice cream doesn't cause sunburn. When someone says "customers who do X buy more," ask whether X causes it or just goes along with it.
Trusting a single number. Ask where it came from, how it's calculated, and what's left out.
Buying tools first. Software helps once you know what you want it to do. Before that, it's an expensive way to feel busy.
Where to go from here
Pick one question that would change a decision. Find the data that could answer it. Answer it as simply as you can, and act on the result. Then pick the next one.
That's most of what "data-driven" means. If you'd like a hand with the first one, or a view on whether your data is ready, we're happy to talk.