Managers do not need to be analysts. They need to be unbluffable - able to tell a solid number from a shaky one, and to ask the question that exposes the difference. That is a much smaller skill set than a statistics qualification, and it has a larger effect on organisational decisions.
Why this matters more than analyst capability
An organisation with excellent analysts and statistically illiterate managers gets worse decisions than one with adequate analysts and literate managers. Why? Because the manager decides what gets analysed, whether the caveats survive into the decision, and whether uncomfortable findings get acted on. Literacy at the top determines whether analysis has any effect at all.
The minimum viable skill set
1. Variation. Know the normal range of your key metrics. Without this you will react to noise, and your team will spend their lives explaining ordinary fluctuation.
2. Denominators. Always ask what a percentage is a percentage of. Thirty seconds, catches a great deal.
3. Missing data. Ask who is not in the dataset. Non-respondents, lapsed customers, rejected candidates - the excluded group frequently holds the answer.
4. Averages and spread. Know that an average conceals the distribution, and ask for a percentile whenever the tail matters (complaints, delays, costs).
5. Correlation versus causation. Specifically: before approving spend based on a relationship, ask what else could explain it.
6. Significance in plain terms. "Is this bigger than the normal wobble?" plus "how big is the effect and is it worth acting on?"
7. Confidence ranges. Insist on them in every forecast and every estimate. A single number in a board pack is a warning sign, not a reassurance.
The five questions to ask in any meeting
- What is the sample size, and who is missing?
- How much does that metric normally move anyway?
- Compared with what - last year, target, the rest of the market?
- What else could explain this?
- What is the range around that estimate?
None require technical knowledge. All of them regularly change conclusions, and asking them consistently trains your organisation faster than any policy.
What to model as a leader
Two behaviours matter disproportionately. First, ask "how do we know?" about your own claims as well as others'. Second, change your stated view publicly when the evidence goes against you - until people see that happen, they will keep bringing you confirmation rather than analysis. This is the foundation of building a data culture.
What you can safely delegate
The methods. You do not need to know how a t-test works, which regression specification is appropriate, or how to build a forecast model. You need to know what to ask of the person who does, and how to spot when the answer is evasive.
The realistic time investment
A focused applied course plus the habit of asking the five questions is enough for most senior roles - weeks, not years. The return is measured in decisions not taken on the basis of a good fortnight.
The Statistics for Business course at London School of Business UK works well for managers wanting exactly this, and can be delivered to leadership cohorts alongside our executive programmes. Talk to us.