Four Analytics Concepts Every Manager Should Understand
You do not need to be a data scientist to work well with data. But almost every manager today is expected to read reports, question dashboards and make decisions based on numbers. The managers who do this well are not necessarily the most technical. They simply understand a handful of core ideas well enough to ask the right questions.
Here are four analytics concepts that will make you a sharper consumer of data, whether you work in banking, telecoms, FMCG, the public sector or a growing startup.
1. Correlation is not causation
Two things moving together does not mean one causes the other. A retailer might notice that branches with more staff also have higher sales. Does adding staff drive sales? Or do busier branches simply get more staff because they are busy? Or is a third factor, such as location, driving both?
This mistake is everywhere in business reporting. Before acting on a pattern, ask:
- Could the cause run the other way?
- Is there a third factor that could explain both?
- Could this be coincidence, especially if we looked at many measures until one stood out?
When someone presents a relationship in the data, the most useful question you can ask is: “How do we know this causes that?”
2. Experiments and A/B tests
The most reliable way to establish cause and effect is to run an experiment. In its simplest business form, the A/B test, you show one group version A and a similar group version B, then compare results. A bank could send two versions of a loan offer by SMS to randomly selected customers and see which one gets more responses.
Two ideas make experiments trustworthy:
- Randomisation. People should be assigned to groups by chance, not by choice. If your best customers end up in group B, you are measuring the customers, not the offer.
- A control group. You need a comparison group that does not get the change. Without it, you cannot tell whether results came from your action or from something else happening at the same time, like a salary week or a public holiday.
Common mistakes include stopping a test as soon as one version looks ahead, testing too many changes at once and running tests on groups too small to tell you anything. Good experiments take patience, but they replace opinion with evidence.
3. Statistical significance, and why it is not the whole story
When a report says a result is “statistically significant”, it means the difference is unlikely to be due to random chance alone. That is useful, but it is often misunderstood.
- Significant does not mean important. With a large enough sample, a tiny difference, such as a 0.2% rise in click rate, can be statistically significant while being commercially meaningless.
- Not significant does not always mean no effect. A small sample might simply be too small to detect a real difference.
- Always ask about the size of the effect and the cost of acting on it, not just whether it passed a statistical test.
The practical question for a manager is: “Is this difference big enough, and reliable enough, to justify the decision we are about to make?”
4. Regression: what drives what
Regression analysis is a family of techniques for estimating how much one factor influences an outcome while holding others constant. A sales team might use it to estimate how price, promotion spend, season and region each affect volume. An HR team might use it to understand which factors are most linked to staff leaving.
You will rarely run a regression yourself, but you will see its results. When you do, ask:
- Which factors were included, and what important ones might be missing?
- How well does the model actually explain the outcome?
- Is the relationship causal, or only an association? Regression on its own does not prove cause.
- Has the model been tested on data it was not built on?
These questions are equally useful when you are handed the output of an AI or machine learning tool. Many of them rest on the same statistical foundations, and the same caution applies.
Becoming a confident data consumer
The goal is not to master the mathematics. It is to recognise when a number deserves trust, when it needs a second look and when someone is drawing a bigger conclusion than the data supports. Managers who can do this make better decisions and earn more credibility with analysts and leadership alike.
If you want to build these skills properly, the Entry Level Analyst Program covers data analysis from the ground up, and the GAA Mini-MBA includes data and decision making alongside strategy and finance. For business analysts, CBAP preparation builds the wider analysis toolkit.
FAQs: analytics for managers
Do managers need to know statistics? Not in depth. Managers need enough understanding to question data, spot common errors and judge whether evidence supports a decision. The four concepts above cover most everyday situations.
What is the difference between an A/B test and a survey? A survey tells you what people say. An A/B test shows you what people do when given different options, which is usually more reliable for business decisions.
How can I check if a dashboard is telling the truth? Ask where the data comes from, how each measure is defined, what period it covers and whether anything unusual happened in that period. Then ask whether the conclusion depends on correlation alone.