ChatGPT for Microsoft 365 Users Smarter Emails, Documents, and Spreadsheets

Introduction

If you use Microsoft 365 every day, you probably do the same types of work again and again:

• writing and replying to emails
• turning rough notes into clear documents
• fixing wording, tone, and formatting
• cleaning up spreadsheets and building formulas
• summarising meetings and pulling out actions

ChatGPT can help with all of that. The key is using it like a thinking and drafting tool, not like a magic button that produces perfect work.

This article is written for everyday Microsoft 365 users. It focuses on practical habits and ready-to-use prompts that help you get better results, while staying sensible about privacy and accuracy.

1) What ChatGPT is best at (and what it’s not)

ChatGPT is great for:

• first drafts (emails, reports, policies, guides)
• rewriting (clearer, shorter, more professional, more friendly)
• summaries (notes → bullet points, decisions, actions)
• structures (headings, checklists, templates)
• ideas (options, pros/cons, risks, next steps)
• explaining (formulas, concepts, steps)

ChatGPT is not great for:

• facts you haven’t provided (it may guess or invent)
• final numbers (you must still check calculations)
• anything that needs access to your files (unless you use an approved, connected setup)
• sensitive decisions about people (you need formal review and accountability)

A safe mindset is this:

ChatGPT can give you a strong draft. You still own the final answer.

2) A simple rule for privacy (before you paste anything)

Before you put text into ChatGPT, ask:

“If this text leaked, would it cause a problem?”

If the answer is yes (or maybe), don’t paste it. Instead:

• remove names, email addresses, phone numbers, job titles, IDs, and account details
• replace them with placeholders (Person A, Client, Supplier, Manager)
• summarise the situation without copying the original text
• keep it “general” and ask for a template or approach

Quick traffic-light guide

• Green: public info, generic templates, your own writing with no sensitive details
• Amber: internal info, meeting notes, drafts that mention non-sensitive company details
• Red: personal data, HR matters, client confidential info, financial account details, passwords, security info

If your organisation has an approved AI tool or policy, follow it. If it doesn’t, keep your use to Green and careful Amber until rules are clear.

3) The secret to better results: give context, constraints, and a goal

Most disappointing ChatGPT outputs come from prompts that are too vague.

Instead of:

“Write an email to a customer.”

Try:

“Write a polite email to a customer who hasn’t replied. Keep it under 120 words. Friendly tone. Ask for an update and offer two time slots for a call. UK English.”

A strong prompt usually includes:

1. Goal: what you want to achieve
2. Audience: who it’s for
3. Tone: friendly, firm, formal, upbeat
4. Constraints: word limit, structure, bullet points, include X, avoid Y
5. Your input: the key facts you want included (without sensitive data)

A simple prompt formula you can reuse

Copy and paste this and fill in the brackets:

Act as a helpful workplace assistant.
Task: [what you want]
Audience: [who it’s for]
Tone: [tone]
Must include: [facts / points]
Must avoid: [words / promises / sensitive details]
Format: [bullets / table / headings]
Ask me up to 3 questions if anything important is missing.

4) Smarter emails in Outlook: drafts, replies, and follow-ups

Email is where ChatGPT often saves the most time. The trick is to use it for the parts that slow you down:

• starting from a blank page
• finding the right tone
• being concise
• structuring a clear request

A) Write a clear email from bullet points

Prompt:

Turn these bullet points into a professional email in UK English. Keep it under 150 words. Add a clear call to action at the end.
Bullets:

• [point 1]
• [point 2]
• [point 3]

B) Reply politely to a difficult message (without escalating)

Prompt:

Draft a calm reply to a frustrated customer. Acknowledge their frustration, apologise for the inconvenience, explain what we can do next, and avoid blame. Keep it friendly and under 170 words.
Situation summary: [short summary, no private data]

C) Ask for a decision (and make it easy to answer)

Prompt:

Write a short email asking for a decision. Give two options, each with a clear outcome. Include a deadline. Keep it polite and direct.

D) Follow up without sounding pushy

Prompt:

Write a friendly follow-up email. Assume the person is busy. Keep it under 120 words. Include the original request in one sentence and ask for a quick update.

E) Turn a messy email into a clean summary and next steps

This is great when you’re copied into a long thread.

Prompt:

Summarise this email thread into:

1. What’s been agreed
2. What’s still open
3. Actions (with owners)
4. Deadlines
   Here’s the text (remove any names first): [paste]

Email quality checklist (30 seconds)

Before you send anything drafted by ChatGPT:

• Did it include anything you shouldn’t share?
• Does it make any promises you can’t keep?
• Is the tone right for the relationship?
• Is the call to action clear?

5) Better documents in Word: structure, clarity, and consistency

ChatGPT is excellent at turning rough thinking into a clean document. It can:

• propose a structure with headings
• rewrite paragraphs in plain English
• tighten wording
• create checklists and templates

A) Turn rough notes into a structured document

Prompt:

Turn these notes into a well-structured document with headings and bullet points. Use plain English (8th grade reading level). Keep it practical.
Notes: [paste]

B) Make a document clearer (without changing meaning)

Prompt:

Rewrite this text to be clearer and easier to read. Keep the meaning the same. Use short sentences and plain UK English.
Text: [paste]

C) Create a standard operating procedure (SOP)

Prompt:

Create an SOP for this process. Include: purpose, scope, roles, step-by-step instructions, exceptions, and a quick checklist at the end.
Process: [describe the process]

D) Improve tone: more confident, more polite, or more neutral

Prompt:

Rewrite this to sound [confident but polite / neutral / friendly and warm]. Avoid buzzwords. Keep it direct.
Text: [paste]

E) Build a reusable template (policies, reports, meeting notes)

Prompt:

Create a one-page template for [weekly project update / meeting notes / incident report]. Include sections that prompt people to write the right information. Keep it simple.

Document habit that makes a big difference

Ask ChatGPT to add a “What to check before publishing” section. It helps you catch gaps.

Prompt:

Add a short “Checks before publishing” list for this document.

6) Smarter spreadsheets in Excel: ideas, formulas, and cleanup

ChatGPT won’t replace Excel, but it can make you much faster in Excel by helping you:

• choose the right approach
• write and explain formulas
• troubleshoot errors
• suggest ways to clean and reshape data

The key is to give it enough detail about your spreadsheet without sharing sensitive data.

A) Ask for the right formula (with your layout)

Provide column names, not real data.

Prompt:

I’m using Excel. I have a table with these columns:

• A: Date
• B: ClientName
• C: Amount
• D: Status
  I want a formula that returns the total Amount for a given ClientName where Status = “Paid”.
  Please give the formula and explain it.

B) Troubleshoot a formula error

Prompt:

This formula returns #VALUE!. Explain the most likely reasons and how to fix them. Also give a corrected version if you can.
Formula: [paste]

C) Turn a requirement into steps (best for complex tasks)

Instead of asking for one big formula, ask for a method.

Prompt:

I need to: [describe the end goal].
Suggest two ways to do this in Excel:

1. using formulas
2. using PivotTables (and Power Query if helpful)
   Explain which is best and why.

D) Cleaning messy data (names, dates, duplicates)

Prompt:

I have messy data with: extra spaces, inconsistent date formats, and duplicated rows.
Give me a step-by-step cleanup plan using Excel features (Power Query if useful). Keep it beginner-friendly.

E) Explain a function in plain English

Prompt:

Explain how XLOOKUP works in plain English, with a simple example using a small table.

Spreadsheet safety checklist

Before you trust any AI-generated formula or method:

• test on a copy of the file
• check results with a small sample you can verify
• sanity check totals (do they roughly match what you expect?)
• don’t assume the first answer is correct

7) ChatGPT + Microsoft 365 workflows that actually work

This is where ChatGPT shines: not as a “do everything” tool, but as the helper that moves your work from rough to ready.

Workflow 1: Meeting notes → summary → actions → email

1. You write rough notes (even messy)
2. ChatGPT turns them into:

• a short summary
• action list with owners
• decisions and deadlines

3. You paste the final actions into your email or Teams message

Prompt:

Turn these notes into: Summary, Decisions, Actions (with owners), Deadlines, and Risks. Use bullet points.
Notes: [paste, remove names if needed]

Workflow 2: Customer enquiry → response options → final email

1. You summarise the situation in 4 to 6 lines
2. ChatGPT writes 2 or 3 versions
3. You choose the best tone and adjust details

Prompt:

Draft 3 reply options:

1. very concise
2. warmer and more detailed
3. firm but polite
   Situation: [summary]

Workflow 3: Spreadsheet requirement → approach → build → explain

1. You describe the goal and columns
2. ChatGPT suggests the best method
3. You build it in Excel
4. ChatGPT explains the logic so you can support it later

Prompt:

I need to build this in Excel: [goal].
Here are my columns: [list columns].
Suggest a method and then explain it so I can recreate it.

Workflow 4: Draft document → improve clarity → final checks

1. You write a first draft quickly
2. ChatGPT improves clarity and structure
3. You do a final review for accuracy and tone

Prompt:

Improve this draft for clarity and structure. Keep the meaning the same. Add headings where useful. End with a short checklist of what to verify.
Draft: [paste]

8) “Human in the loop”: the habit that keeps you safe

The biggest risk with AI at work isn’t using it. It’s using it without checking.

Human-in-the-loop simply means:

• a real person reviews the output
• checks facts and numbers
• decides what is acceptable
• takes responsibility for the final version

Use this quick review checklist

Before you send, publish, or rely on AI output:

1. Privacy: Did I include anything sensitive?
2. Accuracy: What facts or numbers must be verified?
3. Tone: Would I be happy if this was forwarded?
4. Promises: Did it commit us to a date, cost, or outcome?
5. Ownership: Am I comfortable putting my name to this?

If the work affects a real person (HR, performance, discipline, hiring), treat it as high risk and use a more formal review process.

9) Prompt packs you can save for quick wins

Here are a few “prompt packs” you can keep in OneNote or a Word file for your team.

Email pack

• “Write a polite follow-up under 120 words…”
• “Rewrite this email to be clearer and more direct…”
• “Give me 3 subject line options for this email…”
• “Turn these bullet points into a professional email…”

Document pack

• “Create an outline with headings for…”
• “Rewrite this to be easier to read (plain English)…”
• “Turn this into an SOP with steps, roles, exceptions…”
• “Create a template for…”

Excel pack

• “Given these columns, suggest the best method…”
• “Explain why this formula returns an error…”
• “Give me a step-by-step plan to clean this dataset…”
• “Explain this function with an example…”

One simple team tip

Ask everyone to add one prompt per week that genuinely helped them. In a month you’ll have a strong internal library.

10) Common mistakes (and how to fix them)

Mistake 1: “Do this for me” prompts

These often produce generic output.

Fix: Give context, audience, and constraints.

Mistake 2: Trusting output because it sounds confident

AI can be wrong in a convincing way.

Fix: Verify facts, and test formulas.

Mistake 3: Pasting sensitive data

This is the quickest way to create risk.

Fix: Anonymise, summarise, or keep it Green.

Mistake 4: Using one prompt and stopping

Great output usually comes from small iterations.

Fix: Ask: “Make it shorter”, “Make it more polite”, “Add a clear call to action”, “Give me 2 options”.

Mistake 5: Forgetting the final human edit

Even when accurate, AI text can sound slightly “off”.

Fix: Read it out loud once. If it doesn’t sound like you, edit it.

Learn how to use AI well at work

If you’d like to use AI more confidently, with practical examples, clear boundaries for privacy, and a focus on accuracy and real workplace value, take a look at ExperTrain’s Artificial Intelligence (AI) courses here:

https://www.expertrain.co.uk/artificial-intelligence-ai

‍