Good analysts are not distinguished by which techniques they know. They are distinguished by the questions they ask before, during and after the analysis. Here is the full sequence.
Before you open the data
1. What decision does this analysis serve? If nobody will act differently based on the answer, stop now.
2. What result would change our mind? Write it down first. This single step prevents most motivated reasoning.
3. What do we already believe, and why? Naming the prior assumption makes it possible to test rather than accidentally confirm.
About the data itself
4. Where did this come from, and how was it collected? Systems, surveys and manual entry each have distinctive failure modes.
5. Who or what is missing? This is the most powerful question in the list. CRM data excludes people who never enquired. Complaint data excludes the dissatisfied customers who silently left. Employee survey data excludes those who did not respond - who are usually the least engaged.
6. What period does it cover, and is that period typical? Any window containing a holiday, a promotion or a system migration needs a caveat.
7. Is the definition of each field what I think it is? "Active customer" means four different things in most companies.
While analysing
8. What does the distribution look like, not just the average? Look at the shape, the spread and the extremes before quoting any mean.
9. How much does this metric normally vary? Without this, no change can be interpreted.
10. Does it hold up when I segment it? A pattern present in the total but absent in every subgroup - or vice versa - usually means something important is being hidden. This is where the genuinely interesting findings live.
Before you present it
11. What else could explain this? List at least three alternatives, including seasonality, a concurrent change, and a data artefact. If you cannot rule them out, say so.
12. How confident am I, and in what range? Give the interval. "Between 3% and 9%" is honest; "6%" implies a precision you do not have.
The question to ask afterwards
Did this analysis actually change what we did? If your analyses never change decisions, either you are answering the wrong questions or the organisation is not ready to use them - and both are worth naming.
Why this list beats technique
Most bad business analysis is technically competent and conceptually broken - the arithmetic is right and the question was wrong. These twelve questions catch that class of error, which is why they are worth more than another statistical method.
The Statistics for Business course at London School of Business UK teaches this framework alongside the techniques. Enquire today.