Copilot in Excel for Finance: Skills, Data Connectors, and Plan with Copilot Explained

Spreadsheets have been the backbone of financial analysis for 40 years. The problem is not Excel — it is the hours spent on the mechanical work: refreshing numbers, reformatting tables, building the same DCF model for the fifth time this quarter. Microsoft has been shipping Copilot into Excel for two years, but the June 2026 update is different. It is the first release that was explicitly designed around how finance teams actually work, validated against Financial Modeling Institute benchmarks.

Here is what changed, what is generally available now, and what your finance team should actually try first.

The design philosophy: transparency over magic

Every AI tool in finance runs into the same wall: the CFO (or the auditor) asks "how did you get that number?" and the model cannot explain its reasoning. Microsoft's answer for Copilot in Excel is Plan with Copilot — before making any changes to a spreadsheet, Copilot shows you exactly what it intends to do, step by step, and waits for your approval.

The second piece is the Show Changes pane, which attributes every edit — whether made by a human or by Copilot — to the specific actor that made it, exactly like tracked changes in Word. An auditor looking at the workbook can see precisely which rows Copilot updated and when.

"The answer alone is not enough. You need to know how you got there." — Brian Jones, VP Excel Product Group, Microsoft

What is generally available now

The six new financial data connectors

Data connectors allow Copilot to pull live financial data directly into Excel without manual CSV exports or copy-paste from terminal screens. These six connectors are now available:

Skills: how to define how Copilot works for your team

Skills are the most significant enterprise feature in this release. A Skill is a markdown file that defines exactly how Copilot should handle a specific recurring task — the steps to follow, the naming conventions to respect, the output format to produce. Once saved to OneDrive, any user on the team can invoke that Skill by name in Copilot.

Example: a DCF modelling skill

# Skill: Build DCF Model

## When to use
When asked to build a DCF (Discounted Cash Flow) valuation model for a target company.

## Steps
1. Create a new sheet named "DCF - [CompanyName]"
2. Pull 5 years of historical EBITDA from the Financials sheet
3. Apply a growth rate from the Assumptions sheet (cell B4)
4. Discount free cash flows at WACC from Assumptions sheet (cell B6)
5. Calculate terminal value using Gordon Growth Model
6. Sum PV of FCFs + terminal value for Enterprise Value

## Output format
- All monetary values in USD millions
- Row headers in column A, years in row 3
- Use the team colour scheme: headers in #1e3a5f, alternating rows in #f0f4f8

## Review
Always show Plan with Copilot before executing. Do not overwrite existing data.

Workbook rules vs personalisation: what is the difference

What IT admins need to do

For most of these features, the answer is: nothing. Copilot in Excel features activate for licensed users automatically. The exceptions:

1. Data connectors require connector approval. Some connectors may need admin consent in the Microsoft 365 admin centre if they use OAuth flows to the external data provider.
2. Custom Skills require users to have OneDrive access and the ability to create files. Standard in most tenants, but check if your OneDrive policies restrict file creation in personal storage.
3. Partner Skills (Q3 2026) will require admin approval via the Microsoft 365 admin centre app governance section.

Checking Copilot licence assignments in Intune / M365 admin centre

Filter by licence type Microsoft 365 Copilot to see which users have the add-on assigned. The finance team needs both a base Microsoft 365 licence (E3 or E5) and the Copilot add-on to access the features in this post.

Practical finance workflows to try first

These are the five prompts Microsoft demonstrated at launch — they work against real workbook data once a data connector is configured:

• Variance analysis: "Run variance analysis on the P&L vs last quarter and write an executive summary in the Summary sheet"
• Rolling forecast: "Update the revenue forecast by rolling forward Q3 actuals from the connected FactSet data"
• Comparable company analysis: "Pull 5-year EBITDA margins for the 8 companies in the Comps sheet from Daloopa and format as a comparison table"
• Acquisition screening: "From the PitchBook connector, find SaaS companies with ARR between $50M and $200M that match our acquisition criteria in column D"
• Earnings summary: "Summarise the earnings sentiment from the last two transcript files in OneDrive and flag any guidance changes vs consensus"

References

• Copilot in Excel: Built for the era of frontier finance — Microsoft 365 Blog
• Microsoft 365 Copilot overview — Microsoft Learn
• Set up Microsoft 365 Copilot — Microsoft Learn
• Excel Blog — Microsoft Tech Community