Level 1 · Frame the price question

Why framing comes first

“Did the price rise hurt sales?” sounds like a question, but it is really a worry. Before you write a query you need to agree what an answer would look like. The method most analysts use is SMART, the same checklist used for goals:

• Specific: which products, which channel, which customers. Here: the 29 products whose price went up on March 1, 2025.
• Measurable: which number moves. Units sold show demand; revenue shows what the business earned. A price rise can lower one and raise the other, so you look at both.
• Achievable: can the data answer it? Yes: every order line in order_items stores the price actually paid, so you can see the change and when it happened.
• Relevant: what decision does it feed? Thomas will keep, roll back or extend the rise.
• Time-bound: which period, and by when. March to August 2025, answer by next Friday.

If you cannot write the question in one sentence that contains all five, it is not ready yet.

Compared with what?

The part that turns a description into an analysis is the comparison. Two are needed here:

• The same months a year earlier. Home goods sell more in November and December. Comparing the six months before March with the six months after puts the holiday peak in one period and spring in the other, so almost every product looks worse after March.
• Products whose price did not change. If the whole store grew 10% this spring, a raised product that grew 10% did not suffer. The unchanged products are a control group: they show what would have happened without the rise.

There is a third trap: products launched during 2024 had only part of a year of sales last spring, so they grow fast whatever their price. A fair comparison uses like-for-like products, the ones on sale in both periods. In this case that means products launched before January 1, 2024.

What the plan looks like

Write the plan down before you start; it doubles as the method section of your write-up:

Step	What you do	Output
Check	Find the products whose price changed and when	a list, plus proof of the switch date
Analyse	Units and revenue, Mar–Aug 2024 vs 2025, raised vs unchanged, like for like	a 2-row table
Show	One chart: units change by group	a column chart
Recommend	What it means for the decision, with caveats	three or four sentences

Interview note

“How would you measure the impact of a price change?” is a classic product and analyst interview question. A strong answer names the metric (units and revenue, sometimes margin), the comparison (same period last year plus a control group), the like-for-like filter, and the limit: without a randomised test you can show an association, not prove the cause.