Purpose
Use this lesson to frame an analysis around a business decision, metric, comparison, and time period. The aim is to produce work that a colleague can review and act on, not merely text that sounds convincing.
Lesson Explanation
Scenario: Sales fell last month and leadership wants an explanation.
Better instruction: Analyze month-over-month sales by product and region; identify the largest absolute and percentage changes; separate observed patterns from possible explanations.
“Analyze this data” is too broad. Identify the decision, metric definition, comparison period, segmentation, and distinction between facts and hypotheses. This keeps the output from presenting correlation as certainty. The analysis specifies products, regions, month-over-month comparison, and separates observations from explanations.
Work in a short cycle: provide the relevant input, state the result and limits, inspect the draft against the acceptance criteria, then revise the instruction when a requirement is missing. Keep the final human decision with the person responsible for the work.
Practice Questions
-
A prompt asks an AI to “analyze this data and tell us what happened” with no further specification. What does this lesson say about this kind of request?
View Answer
“Analyze this data” is too broad – the decision, metric definition, comparison period, and segmentation need to be identified.
-
What does this lesson say the analysis output should separate?
View Answer
Observed patterns from possible explanations – facts from hypotheses.
-
An AI-generated analysis states a correlation between two variables as if it were a proven cause. What problem does this lesson identify with this kind of output?
View Answer
It presents correlation as certainty, rather than keeping facts distinct from hypotheses about causation.
-
Write a prompt for analyzing a drop in website traffic, following this lesson’s guidance on what to specify.
View Answer
Something like: “Analyze month-over-month website traffic by page and traffic source; identify the largest absolute and percentage drops; separate observed patterns from possible explanations.”
-
A prompt for sales analysis specifies the metric (revenue) and comparison period (month-over-month) but doesn’t specify how the data should be segmented. What might this affect?
View Answer
The analysis might miss important patterns that only show up when broken down by category, region, or another relevant segment.
-
Why does this lesson caution against letting an analysis “present correlation as certainty”?
View Answer
Two things being correlated doesn’t prove one caused the other, and presenting this as certain could lead to incorrect business decisions.
-
A completed analysis specifies the products and regions examined, uses a month-over-month comparison, and clearly separates observations from explanations. What does this reflect?
View Answer
A well-specified analysis prompt following this lesson’s full guidance.
-
A prompt asks “why did sales drop?” without defining the comparison period (e.g., versus last month, last year, or launch). What is missing, per this lesson?
View Answer
A defined comparison period.
-
An analysis identifies that two metrics moved together during the same month and presents this as the definitive cause of a sales decline. What does this lesson suggest should have been done differently?
View Answer
The output should have separated this observed correlation from any explanatory hypothesis, rather than presenting the pattern as a proven cause.
-
Why does this lesson recommend identifying “the decision” an analysis needs to support, rather than just requesting general analysis?
View Answer
Defining the decision clarifies what specific question the analysis needs to answer, shaping the metric, comparison period, and segmentation choices appropriately.
-
A prompt requests analysis segmented by both region and product line, with month-over-month comparison, and a request to flag the largest percentage changes specifically. What does this reflect?
View Answer
A specific, well-defined analysis request consistent with this lesson’s guidance.
-
A report says “revenue and marketing spend increased together, so marketing spend caused the revenue increase.” What does this lesson say is wrong with this conclusion?
View Answer
It presents a correlation as a proven cause, when the analysis should have kept the observed pattern separate from unproven explanatory hypotheses.