You can run a business well with five statistical concepts. Each answers a distinct question, and each has a specific way of misleading you if you use it alone.
1. The mean (average)
Answers: what is typical, in total terms? Revenue per customer, cost per order, days to pay an invoice.
The trap: a handful of extreme values drag the mean away from reality. One enterprise client can make your "average deal size" a figure no actual deal has ever matched.
2. The median
Answers: what does the middle of my data look like?
The trap: the median ignores extremes entirely, which is usually the point - but occasionally those extremes are the business. Always look at both. If mean and median are far apart, your data is skewed and you need to segment it before deciding anything.
3. Standard deviation (spread)
Answers: how much do my numbers normally bounce around?
This is the most neglected and most useful of the five. Without a sense of normal variation, you cannot tell a real change from a random one - and you will end up rewarding lucky months and punishing unlucky ones. Know your normal range and most "surprises" stop being surprising.
The trap: it assumes a reasonably symmetric distribution. For heavily skewed data (order values, waiting times), look at percentiles instead.
4. Percentage change - and rate versus count
Answers: how much has this moved, and is it moving relative to a fair base?
The trap: percentages of small numbers are meaningless ("enquiries up 200%" = two became six), and percentage changes on percentages cause endless confusion. Also beware comparing a rate with a count - more complaints is not the same as a higher complaint rate if your customer base grew.
5. Correlation - with the causation caveat
Answers: do these two things move together?
Extremely useful for spotting where to look: ad spend and sales, staffing and delivery times, discounting and margin.
The trap: the famous one. Correlation is not causation, and the most common business error is spending money on the basis of a relationship that a third factor explains. Ice cream sales correlate with sunburn; neither causes the other. When a correlation matters commercially, test it deliberately.
The sixth thing, which is not a statistic
Sample size and how the data was collected. Any of the five above, computed on a biased or tiny sample, produces confident nonsense. Ask where the numbers came from before you ask what they say.
Putting them to work
These five plus a spreadsheet handle most operational decisions. When you are ready to add confidence intervals, hypothesis tests and forecasting, the Statistics for Business course at London School of Business UK covers them applied to business data. Enquire today.