Signing in

You will be sent to MillerKnoll sign-in.

AI

What Does Copilot See?

Before analysis, agree with Copilot on what is in the workbook. Send a setup prompt, paste data, say Go — Copilot reports columns, row count, time period, quality issues, and one business headline you can sanity-check.

Lesson 1

Read first — do not edit yet.

This lesson does not let Copilot change anything. You are calibrating — does its read match what you know is in the file?

Core principles

  1. Format the range as a Table (Insert → Table) before deeper analysis — Copilot reads tables more reliably.
  2. Keep raw data on its own sheet; name it clearly (e.g. Raw Data).
  3. Missing July revenue, TBD in a numeric column, and blank cells are quality signals — not noise to ignore.
  4. The headline Copilot offers is a hypothesis — trace it before you repeat it in a meeting.
  5. If Copilot's column meanings are wrong, fix your understanding in Lesson 3 before correlating anything.

Check yourself

What should Copilot do before you say "Go" in the simple Excel prompt?

Do this in Copilot

Run the simple prompt flow with the sample data or your own sheet. Note one quality issue Copilot caught that you had missed.

Send this in Excel Copilot, paste data into A1 on a sheet named Raw Data, then say: Go. The data is on the active sheet. Proceed with the analysis.

Simple prompt (send first)

I will paste sample data into the active sheet next. When I say "Go", analyze that data and tell me:
1. What columns exist and what each appears to mean
2. How many rows and what time period is covered
3. Two obvious data-quality issues you see
4. One simple business headline a MillerKnoll associate could sanity-check

Rules:
- Do not change anything in the workbook
- Do not analyze until I say "Go"

Do not analyze yet. Reply "Ready — paste your data, then say Go." when you understand these rules.
Open Copilot →
  • Two-message flow
  • Constraints
  • Sample data (paste into A1):
  • Month,Revenue_EUR,Orders,Industry_Shipments,New_Contracts,Notes Jan 2024,1240000,412,8900,38, Feb 2024,1185000,398,8750,41, Mar 2024,1310000,425,9100,,Q1 close Apr 2024,1275000,401,9050,29, May 2024,1340000,430,9200,33, Jun 2024,1290000,415,9150,31, Jul 2024,,408,9000,27,holiday slowdown? Aug 2024,1220000,390,8800,22, Sep 2024,1365000,442,9300,36, Oct 2024,1410000,455,9450,40, Nov 2024,1380000,448,9400,38, Dec 2024,1520000,470,9600,45,strong finish Jan 2025,1280000,405,9100,30, Feb 2025,1190000,392,8950,28, Mar 2025,1330000,418,9250,34, Apr 2025,1305000,410,9180,32, May 2025,1375000,428,9350,35, Jun 2025,TBD,420,9280,33,preliminary

Did you run this in Copilot? Mark complete when you have tried it.

Next lesson: Raw Data Is Sacred →