Analyze Post-Promotion Results From an Exported Spreadsheet With ChatGPT

The tool works best with clear column names and one record per row, which OpenAI's data analysis help page describes.

1. Open your export and save a copy named for the promotion, such as "SPRING-A_results_clean.xlsx".
2. Delete personal columns and any blank rows or subtotal rows.
3. Make the first row a single header row, for example: promo_code, retailer, week, spend, redemptions.
4. Put the plan figure for cost per redemption in your own notes, ready to type into the prompt.

What you should see: A plain table with one header row and one line per retailer and week.

You do not need a special tool. A normal chat can analyze spreadsheets when your plan allows it.

1. Open a new chat at chatgpt.com.
2. Use the attach or plus control beside the message box to add your file.

What you should see: The file name appears above or inside the message box. Troubleshooting: If there is no attach control or the upload is refused, your plan or your workspace settings may not include it. Check OpenAI's help center for current limits, and consider whether a smaller file helps.

Paste a prompt like the one below, with your own plan figure.

1. Say what each column is.
2. Ask for the calculation to be shown.
3. Ask for the three biggest differences and one chart.

Example prompt to copy-paste:

The attached sheet is results for promotion SPRING-A.
Columns: retailer (code), week, spend (a number, no currency symbol), redemptions (a count).
Plan cost per redemption is [plan figure].

1. Calculate cost per redemption (spend divided by redemptions) for each retailer by week, and for each retailer overall.
2. Compare each result with the plan figure and show the difference.
3. List the three largest differences from plan and say which retailer and week each belongs to.
4. Make one chart that shows cost per redemption by retailer and week.
5. Show me the calculation you ran, including the formula and any rows you excluded and why.
Do not guess at missing values. If a row has no redemptions, list it instead of calculating.
Do not explain why a result happened. Only report what the numbers show.

What you should see: A table, a short list of the three largest differences, a chart, and a section showing the code or formula used. Troubleshooting: If a number looks wrong, ask "Which rows did you use for retailer B in week 1?" and compare with the sheet.

This step is not optional. Pick at least three figures, including the one you plan to highlight, and recalculate them in Excel or Sheets by dividing spend by redemptions. If one does not match, find out why before you trust any of them.

Ask for the same table sorted by difference, or for week totals across retailers. Recheck any new figure the same way.