A chatbot is not a data team. It can help you phrase the question and spot a formula. It will also invent a trend if you sound like you want one.
Aside. The robot does not get your badge. Your manager still reads the email with your name on it.
Before you paste
Drop email, phone, address, free-text notes about people. Keep ids, dates, amounts, categories. If the file is someone else’s customers, use the company tool or do not use a model.
Questions that are allowed
- How many rows in category X?
- What is the sum of amount for March? Show the rows you used.
- Which ids appear more than once?
Questions that are not
- “What story should I tell leadership?”
- “Fill the empty revenue cells”
- “Predict next quarter” from twelve rows and a dream
Packet
Use only this table.
If a number is missing, say NEED. Do not impute.
Question: [one question].
Then tell me how to check it in the spreadsheet.
Open the sheet. Recalculate. If the model is off, you stop. You do not ask it to “be more insightful.”
Try this today
One export. One question. One check. That is analytics literacy.
You are done when
You have a number you recomputed, not a paragraph about “hidden stories in the data.”