← Data Analytics from Scratch: SQL, Spreadsheets and Metrics
Lesson
Absolute vs relative numbers in practice
Learner can decide when an absolute number, a rate, or a percentage best answers a question and can spot when a relative number hides a tiny or huge base.
Absolute vs relative: two questions, two answers
Absolute vs relative: two questions, two answers
Absolute numbers count things directly — orders, dollars, users, cases. They answer the question 'How many?' Relative numbers — percentages, rates, per-capita figures — answer 'How intensely?' or 'Compared to what?' Choosing the wrong format creates misleading conclusions, even when the arithmetic is correct.
Why you cannot compare different bases using only absolutes: Store A grew from 50 to 75 orders per day (+25 orders). Store B grew from 5,000 to 5,025 orders per day (+25 orders). In absolute terms the gain looks identical. But Store A grew by 50% while Store B grew by only 0.5%. These are completely different stories — Store A had a breakthrough, Store B barely moved. Relative numbers (rates, percentages) remove the effect of scale and make the comparison honest.
When should you normalise? Comparing teams of different sizes — use output per employee. Comparing cities of different populations — use cases per 10,000 residents. Comparing days with different traffic levels — use conversion rate, not raw click count. Normalisation is just dividing by the right base so that scale no longer distorts the picture.
The base-rate trap: '300% increase' sounds explosive — but if you started with 1 case and now have 4, the absolute change is 3 cases. Context matters enormously. Best practice: always pair a relative figure with the absolute numbers behind it. '4 cases, up from 1 — a 300% increase' is honest. '300% increase!' alone is not. The same logic applies in reverse: a huge absolute number can hide a tiny relative share.
Lesson notes
Absolute vs relative: two questions, two answers
Absolute numbers count things directly — orders, dollars, users, cases. They answer the question 'How many?' Relative numbers — percentages, rates, per-capita figures — answer 'How intensely?' or 'Compared to what?' Choosing the wrong format creates misleading conclusions, even when the arithmetic is correct.
Why you cannot compare different bases using only absolutes: Store A grew from 50 to 75 orders per day (+25 orders). Store B grew from 5,000 to 5,025 orders per day (+25 orders). In absolute terms the gain looks identical. But Store A grew by 50% while Store B grew by only 0.5%. These are completely different stories — Store A had a breakthrough, Store B barely moved. Relative numbers (rates, percentages) remove the effect of scale and make the comparison honest.
When should you normalise? Comparing teams of different sizes — use output per employee. Comparing cities of different populations — use cases per 10,000 residents. Comparing days with different traffic levels — use conversion rate, not raw click count. Normalisation is just dividing by the right base so that scale no longer distorts the picture.
The base-rate trap: '300% increase' sounds explosive — but if you started with 1 case and now have 4, the absolute change is 3 cases. Context matters enormously. Best practice: always pair a relative figure with the absolute numbers behind it. '4 cases, up from 1 — a 300% increase' is honest. '300% increase!' alone is not. The same logic applies in reverse: a huge absolute number can hide a tiny relative share.