Thinking Like an Analyst

Analysis fails at the question more often than at the math. 'Look at this data and find something interesting' produces nothing; 'did the new opening hours change weekday sales?' produces an answer. The analyst's first skill is converting fuzzy curiosity into questions data can actually answer.

Key Concepts

• Every analysis starts with a decision someone must make — find it before touching data
• Precise questions name a metric, a comparison and a period: 'average daily sales, before vs after March, weekdays only'
• Know your data's grain: what does one row represent? One sale? One day? One customer? Everything depends on this
• Distinguish description (what happened), diagnosis (why), prediction (what next) and prescription (what to do)
• Beware convenient data: what you have is rarely all that matters — note what is missing
• Write the question down before analysing; drifting questions produce cherry-picked answers

In Practice

A practical framing tool: complete the sentence 'We will decide ___ based on whether ___ is above/below ___.' If you cannot complete it, the request is not yet an analysis — it is a fishing trip.

Try It Yourself

Take one vague question from your work or studies ('are students doing okay?') and convert it into two precise, answerable questions with metric, comparison and period. Note which decision each would inform.