thinkhard.SI

Prompt repository Data & spreadsheets

Break down sales data by product, region and rep

You need sales performance sliced and explained so you know what is driving the numbers.

Registry no.
DAT-017
Collection
Data & spreadsheets
Laid out for
Any chat model
Score
10 out of 10
Blanks to fill
6
Added

Data & spreadsheetsAny chat

The prompt

Score 10 of 10, Excellent

Use it when You need sales performance sliced and explained so you know what is driving the numbers.

  • {{data}}
  • {{column_notes}}
  • {{periods}}
  • {{targets}}
  • {{known_events}}
  • {{audience}}
Full prompt and why it works
Prompt
Sales data (pasted below or attached):
"""
{{data}}
"""

What the columns mean (whether amounts are before or after discounts and returns, the currency, and which date each row uses):
"""
{{column_notes}}
"""

Periods to compare: {{periods}}
Targets or quotas, if any: {{targets}}
Known events in the period (price changes, new reps, lost accounts, promotions): {{known_events}}
Who it is for: {{audience}}

Task: Break down this sales data by product, region and rep, and explain what is driving the change between the periods, for {{audience}}. The goal is to show what moved the total, with every figure traceable to the data.

Deliver:
1. Headline: total sales in each period and the change in amount and percent, with the calculation on one line, for example "(148,200 - 131,500) / 131,500 = 12.7%".
2. Three tables, for product, region and rep, each with the columns Name, Sales (first period), Sales (second period), Change, Change %, Share of total change, sorted by the size of the change. Check that each table's changes add up to the total change, and show the sum.
3. Drivers: where the columns allow, split the change into volume (more units or customers), price, and mix (a different blend of products). Say plainly if the data cannot support this split.
4. Concentration: how much of the change comes from the top 3 items in each table, and any single deal or customer large enough to distort the picture, with totals shown with and without it.
5. Points to discuss: up to 5 bullets, each tied to a number above.

Requirements:
- Use only the data and events above. Where an event lines up with a change, say they coincide; do not claim it caused the change.
- Compare reps against their targets where given, and note differences in territory or account mix if the data shows them. A ranking by total sales is not a judgment of anyone's performance and should not read as one; the manager draws conclusions about individuals.
- Show every percentage with its underlying amounts.
- Keep the text outside the tables under 400 words.

If the periods are uneven (for example, one has more selling days) or a column needed for a table is missing, say so before the tables. Rows with no product, region or rep go in a "Not recorded" line, counted, not dropped.

Variables

{{data}}
The sales rows with a header row, at least date, product, region, rep and amount, pasted as CSV or attached.
{{column_notes}}
What each column means: gross or net of discounts and returns, the currency, and whether the date is order, invoice or payment date.
{{periods}}
The two periods to compare, for example "Q3 this year against Q3 last year".
{{targets}}
Targets or quotas by rep, region or product, or "none".
{{known_events}}
Anything that happened in the period that could affect sales. Write "none known" if nothing.
{{audience}}
Who will read the breakdown, for example "the sales director" or "the regional managers' meeting".

Why it works

  • Format. A share of total change column shows which products, regions or reps actually explain the movement, and each table's changes must add up to the total.
  • What to avoid. Events that line up with a change are described as coinciding, not as causes, and a ranking of reps is not presented as a judgment of their performance.
  • If unsure. Uneven periods and rows with no product, region or rep are flagged and counted instead of being silently dropped.

How to use it

  1. Press Fill in variables and type your details into each blank. What you type stays in this browser.
  2. Copy the filled prompt into the AI tool you already use, such as ChatGPT, Claude or Gemini, or open it in the Grader or the Prompt Tester here.
  3. Before you rely on it, try it on three or four real examples. The Prompt Tester runs them all at once and shows which ones go wrong.