Flow in Finance

Is Microsoft Copilot the Closest Thing to a CFO’s Inner Peace

I still remember having budget models where my manager hardcoded a number just to meet a meeting deadline, only for that number to sit untouched for months because no one knew it was there… until the board asked a question we couldn’t answer cleanly.

Another time, during OP2 (Amazon’s year-end budget process), it took my team three full days to reconcile data from the OP1 doc (our initial written budget plan), the so-called “final” version of the model, and what was in the system, before we could even start updating OP2.

These are exactly the kinds of painful experiences we’d all like to erase from memory.

Traditional financial forecasting and analysis, often reliant on manual data handling and static models, is time-consuming, error-prone, and all too easy to miss critical insights.

Thanks to Microsoft’s Copilot, we might finally be entering a new era of flow and maybe even a little inner peace when it comes to financial planning and operational excellence.

Copilot really empowers finance professionals, from executives and financial analysts to collection managers and auditors, by automating routine and time-consuming tasks, allowing teams to focus on high-impact decision-making and value-added analysis.

Here are few things Copilot can do:¹

Accelerated Business Case Development: Copilot can significantly speed up the creation of business cases. It assists with integrating formulas, generating charts, testing various scenarios, and identifying the largest cost variances. Finance professionals can leverage Copilot to efficiently pull historical sales and revenue data from financial systems using secure plugins built with Copilot Studio, and then transform project briefs into polished executive presentations.

Optimized Collections and Cash Flow: Copilot enhances collections processes by organizing information from past customer interactions, drafting clear communications outlining collection procedures, and generating follow-up messages regarding outstanding balances or payment plans. It can also analyze different payment methods for effectiveness and help identify customers with outstanding invoices, contributing to improved cash flow and reduced days sales outstanding.

Reduced Operational Spend: Finance teams, often operating as cost centers, constantly face the challenge of achieving more with fewer resources. Copilot automates simple tasks that might typically be outsourced or performed by low-level employees, thus reducing departmental spending. It streamlines drafting approval requests, status emails, and supplier communications, summarizes stakeholder meetings, and assists with manual tasks, enabling more processes to be completed internally.

Enhanced Financial System Utility: Copilot can be extended into existing ERP systems and other financial processes, serving as a powerful layer to bring disparate data together and accelerate decision-making without the need for expensive and disruptive updates to current financial infrastructure.

Robust Risk Management and Compliance: In today’s volatile markets, timely insights into financial risk are paramount. Copilot supports better risk mitigation and compliance by improving visibility across operations. It can help detect potential risks stemming from regulatory changes, automate financial query management through AI-powered chatbots, and rapidly update accounting guidance to reflect new rules. Furthermore, it enhances scenario planning, allowing finance teams to respond faster and more strategically to market changes.

The finance function is the compass of any organization, and with Copilot, finance leaders gain an unprecedented level of intelligence and automation.

From a governance perspective, Microsoft Copilot follows the same strict privacy, security, and compliance standards as Microsoft 365, covering GDPR and the EU Data Boundary. Your prompts, responses, and any data pulled through Microsoft Graph stay inside your organization’s Microsoft 365 environment and are never used to train the underlying AI models, keeping your proprietary information secure and confidential.

To learn more about how Microsoft Copilot can transform your financial operations and empower your team, join us at the AI Finance Club, an AI-first community for finance professionals offering diverse AI learning paths, live monthly workshops, expert network access, monthly AI news reports, and process deep dives.

FAQ

1. Which finance roles can benefit from Microsoft Copilot? Copilot is designed to assist a wide range of finance professionals, including:

• Executives.

• Accountants, Auditors, Analysts.

• Operations, Procurement professionals.

• Collection Managers.

• Financial Analysts.

• Budget Analysts, Financial Planners, Treasury Managers, Risk Managers, Controllers, Data Managers, and Strategic Planners.

• Specialized roles such as Income Tax Compliance Managers (e.g., in extracting information from tax returns or summarizing email threads related to returns).

• Audit, Risk, and Compliance (ARC) Data Solution Managers (e.g., in generating SQL queries, automating alerts, or creating project monitoring pages).

• Vendor Engagement Managers (e.g., in managing vendor compliance data, drafting reports, and leading meetings).

2. What specific tasks can Copilot automate or speed up within Excel for finance? Within Excel, Copilot significantly enhances productivity for finance tasks by:

• Generating insights from existing data, helping to identify information that might have been missed.

• Applying complex conditional formatting to highlight specific data points, such as sales for a particular plan with revenue above a certain threshold.

• Suggesting and creating new formula columns, including calculations for new Key Performance Indicators (KPIs) like revenue per active user.

• Writing complex Excel formulas, even for multi-month projections (e.g., showing monthly revenue for the next 12 months), saving significant time.

• Building mini financial models from a blank page, including structures for headcount, salary costs, tax costs, and bonuses for specified periods, complete with formulas and assumptions.

• Automating data preparation by converting ranges to Excel tables, removing duplicate entries, trimming spaces, and validating data types.

• Drafting multi-year income statements and integrating balance sheet and cash flow statements by linking financial results and computing cash flow from operations.

• Streamlining reconciliation by comparing ledgers, flagging unmatched items, and suggesting adjusting entries, such as generating reports for cash accounts with discrepancies.

• Automating variance commentary by compiling key drivers, calculating variances, and drafting executive-ready analysis.

• Simplifying scenario analysis by building best-case, base-case, and worst-case scenarios, and generating data tables and tornado charts for sensitivity testing.

• Integrating with Python to run advanced analytics directly within Excel, such as Monte Carlo simulations, Value at Risk calculations, and training custom machine-learning models for forecasting.

• Creating data visualizations like scatter charts, count plots, and dashboards from your data.

• Sorting and filtering data based on specified criteria.

• Building machine learning models directly within Excel to predict financial metrics like salary information, leveraging advanced reasoning capabilities.

• Handling text data, enabling tasks like generating word clouds or performing sentiment analysis on customer reviews.

3. How does Copilot interact with other Microsoft 365 applications to support finance workflows? Copilot extends its capabilities across the Microsoft 365 suite, including:

• Microsoft Word: Helps draft various communications, such as approval requests, status emails, supplier communications, and summaries of collections cycles for legal teams. It can also revise content, improve clarity, and create project briefs from existing documents or emails.

• Microsoft PowerPoint: Can transform Word documents or project briefs into new presentations, organize slides, improve slide layouts, and incorporate financial information directly.

• Microsoft Outlook: Assists in managing email communications, drafting follow-up emails, summarizing long email threads, and coaching message tone for clarity and effectiveness.

• Microsoft Teams: Helps summarize stakeholder meetings (e.g., around budget status and approvals), recaps discussions, identifies action items, and assists in preparing for and scheduling follow-up meetings. It can also be used during meetings to keep track of discussions.

• Microsoft SharePoint: Enables users to quickly get answers about previous tax returns or other documents without opening large files.

• Microsoft Power Automate: Supports the creation of automated flows, such as sending alerts about Power BI reporting overloads, analyzing automation activities, or verifying tax return data using OCR models.

• Microsoft Power BI: Assists in developing reporting for stakeholders, including creating project monitoring pages with scheduled hours, resource availability, and identifying gaps or overloads.

• Microsoft Power Apps: Helps in building applications, such as those designed to collect insights from files or text input boxes, by suggesting relevant fields and creating app screens.

• Copilot Studio: Allows organizations to build custom plugins for secure data retrieval from existing Enterprise Resource Planning (ERP), Customer Relationship Management (CRM), and other finance systems. It can also be used to create AI agents, for instance, to identify customers with outstanding invoices.

• Copilot Chat: Functions as a general AI assistant to manage tasks, create meeting agendas from chat history, and respond to general queries.

4. What is required to use Microsoft Copilot? To use Microsoft Copilot, you generally need a Microsoft 365 subscription that includes Copilot access. This subscription provides Copilot functionalities within Excel, PowerPoint, Outlook, and Word. Specifically, for Copilot to run in Excel, the autosave feature must be enabled for the file you are working on.

5. How difficult is it to implement Copilot within our existing finance workflows? Implementing Microsoft Copilot is designed to be straightforward. Since Copilot is embedded directly into the Microsoft apps your teams already use daily, there's no complex rollout or steep learning curve typically associated with new software solutions. The goal is to integrate AI capabilities seamlessly into familiar platforms, enabling immediate productivity gains.

6. Can Copilot integrate with our existing ERP, CRM, and other financial systems? Yes, Copilot is designed for extensibility and can integrate with existing ERP systems, CRM platforms, and other financial processes. It can bring data together from these disparate sources to speed up decision-making without necessitating expensive updates or overhauls to your current financial systems. This integration is facilitated through the use of plugins, which can be custom-built with Copilot Studio for secure data retrieval from your existing systems.

7. What kind of return on investment (ROI) can we expect from using Copilot in finance? Organizations can expect significant ROI from deploying Copilot in finance through various benefits:

• Increased Productivity: Automating mundane and repetitive tasks frees up finance professionals. A leading Australian life insurer, for example, reported staff saving an average of six hours per week on report preparation and reconciliation within three months of deploying Copilot.

• Cost Reduction: Copilot helps reduce departmental spending by automating tasks that might otherwise be outsourced or require lower-level employees, enabling more work to be done internally. It also helps avoid costly financial software upgrades by integrating with existing systems.

• Improved Accuracy: The AI-guided experiences and ability to analyze vast data enhance the accuracy of financial projections. The Australian insurer also noted a 20% uplift in forecasting accuracy.

• Faster Decision-Making: By accelerating data analysis, business case development, and insights across procurement and spend data, Copilot enables quicker and more informed decisions.

• Enhanced Strategic Focus: With routine tasks automated, finance teams can dedicate more time to value-added analysis, proactive audits, and strategic financial reporting, driving greater business impact.

8. How does Copilot handle building and manipulating complex financial models? Copilot significantly accelerates the construction and modification of financial models through natural language interaction. Key capabilities include:

• Automated Tasks: It automates routine tasks such as data reconciliation, variance reporting, and scenario generation within models.

• Natural Language to Excel: Users can type plain-English requests (e.g., "Model how a 2 percent increase in our sales growth rate affects gross margin over the next three years"), and Copilot translates these into Excel formulas, data tables, and visualizations. It can adjust assumptions, rebuild linked schedules, and even annotate the sheet with commentary.

• Data Preparation: Before modeling, Copilot can standardize data by converting ranges to tables, removing duplicates, trimming spaces, and validating data types to ensure reliable inputs.

• Statement Generation: It can draft multi-year income statements from assumptions, and then link operating results to the balance sheet and cash flow statements, ensuring consistency across all financial statements.

• Scenario Analysis: Copilot simplifies "what-if" analysis by building best-case, base-case, and worst-case scenarios, and generating data tables and tornado charts to show sensitivity to various assumptions.

• Python Integration: Excel’s Python integration is fully accessible through Copilot, enabling advanced analytics like Monte Carlo simulations, Value at Risk calculations, and the training of custom machine-learning models for forecasting directly within the worksheet, with all outputs appearing alongside your data.

9. How does Microsoft Copilot ensure data privacy and security for our sensitive financial data? Microsoft Copilot is built on a foundation of enterprise-grade security and adheres to stringent privacy and compliance commitments:

• Compliance: It is fully compliant with existing privacy, security, and compliance commitments for Microsoft 365 commercial customers, including the General Data Protection Regulation (GDPR) and European Union (EU) Data Boundary, as well as standards like ISO/IEC 27018.

• Data Residency: All prompts, responses, and data accessed through Microsoft Graph remain within the Microsoft 365 service boundary. This means your data stays within your Microsoft 365 environment. For EU users, traffic stays within the EU Data Boundary, and Microsoft Advanced Data Residency (ADR) and Multi-Geo Capabilities offerings include data residency commitments for Copilot customers.

• No Model Training: Crucially, your prompts, responses, and the proprietary data accessed through Microsoft Graph are not used to train the foundation Large Language Models (LLMs) that power Microsoft 365 Copilot. This commitment ensures your confidential data remains private and is not used to improve models for other customers.

• Access Control: Copilot only surfaces organizational data to which individual users already have at least view permissions. It strictly honors the user identity-based access boundary, leveraging the same underlying controls for data access used in other Microsoft 365 services like SharePoint. This helps prevent unintentional data leakage. It also honors usage rights for data encrypted by Microsoft Purview Information Protection.

• Encryption and Isolation: Microsoft employs rigorous physical security and a multi-layered encryption strategy, including BitLocker, per-file encryption, Transport Layer Security (TLS), and Internet Protocol Security (IPsec) to protect customer content at rest and in transit. Logical isolation of customer content within each tenant is achieved through Microsoft Entra authorization and role-based access control.

• Azure OpenAI Services: Microsoft 365 Copilot uses Azure OpenAI services for processing, which do not cache customer content. It does not use OpenAI’s publicly available services.

• Content Protections: Copilot operates with multiple built-in protections, including content filtering systems to block harmful content (e.g., hate, sexual, violence, self-harm categories), detection for protected materials (like copyrighted text or code subject to licensing restrictions), and proprietary classifiers to block prompt injections (jailbreak attacks). Abuse monitoring, which includes human review, is opted out for Microsoft 365 Copilot services.

10. Who owns the content generated by Microsoft Copilot? Microsoft does not claim ownership of the output generated by the Copilot service. However, it is important to understand that generative AI systems, due to their nature, may produce similar or substantially similar responses to similar prompts or queries from multiple customers. Consequently, multiple customers may have or claim rights in such content. To address potential concerns, Microsoft offers a Copilot Copyright Commitment for its commercial customers. If a third party sues a commercial customer for copyright infringement specifically for using Microsoft’s Copilots or the output they generate, Microsoft will defend the customer and pay the amount of any adverse judgments or settlements that result from the lawsuit, provided the customer used the guardrails and content filters built into the products.

11. How reliable is the content and analysis provided by Copilot? While Microsoft continues to enhance Copilot’s responses, the content generated by generative AI is not guaranteed to be 100% accurate. Users should always apply judgment and review outputs before sharing them. Copilot is designed to provide useful drafts and summaries to help users work more efficiently, not to fully automate tasks without human oversight. The system makes its decision process transparent by noting limitations, linking to sources, and prompting users to review, fact-check, and adjust content based on their subject-matter expertise. Maintaining a Human-in-the-Loop is essential when using Copilot.

12. Can we track and audit actions performed by Copilot? Yes, every action performed by Copilot is logged and tracked within an Insights pane, providing a comprehensive audit trail. This log includes:

• Prompt History: Captures your exact instructions with timestamps.

• Formula Changes: Records before-and-after cell formulas.

• Version Snapshots: Allows for rolling back to prior model states. This capability enables easy review of edits, supports audit compliance, and helps in documenting change rationales for financial models and other documents.

13. Can't my team just use general AI chatbots like ChatGPT for Excel formulas and analysis? While general AI chatbots like ChatGPT can be effective for exploring ideas and generating complex formulas, Microsoft Copilot offers distinct advantages:

• Native Integration: Copilot is seamlessly integrated directly into Microsoft 365 applications like Excel, Word, and PowerPoint, providing a unified and intuitive user experience without the need to switch between applications.

• Data Security and Privacy: Copilot operates within your company's private Microsoft 365 tenant. This means it can be used safely with confidential company data, as prompts, responses, and data accessed are not used to train public LLMs, and remain within Microsoft's secure service boundary. This is a critical differentiator for sensitive financial data compared to general-purpose chatbots.

• Direct Execution and Output: Copilot can directly run code (e.g., Python within Excel) and display the output within the application itself, streamlining workflows and reducing the need for manual copy-pasting.

• Specialized Agents: Copilot Studio allows for the creation of specialized AI agents tailored to specific financial modeling, data retrieval, or analysis needs, providing more targeted and efficient assistance.

14. Will Copilot replace human finance professionals? No, Copilot is fundamentally designed to be an AI assistant that augments and amplifies human capabilities, not to replace finance professionals. Its purpose is to empower finance teams by:

• Automating repetitive and time-consuming tasks, thereby freeing up professionals to engage in more strategic, analytical, and value-added activities.

• Enhancing productivity and accuracy, enabling finance teams to achieve more with existing resources rather than necessitating reductions in headcount.

• Providing useful drafts and summaries, which serve as a starting point for human review and refinement, ensuring that critical human judgment and expertise remain central to financial processes. Copilot serves as a valuable tool that allows finance professionals to focus on the higher-level, strategic aspects of their roles, transforming their day-to-day work for greater impact.

BONUS: 5 Use Cases

Learn more and join the AI Finance Club here.

1. Generating Data Insights and Dashboards

• Purpose: To quickly identify hidden patterns and trends in large datasets and rapidly build dashboards for preliminary analytics.

• Prompts to try:

"Show me insights on the revenue".

"Can I see other insights" (repeatedly to generate more insights).

"Add all insights to the grid" (to create a dashboard).

"Could you please create a dashboard for me including five data visualizations".

"Show insights in charts".

2. Performing Conditional Formatting with Formulas

• Purpose: To visually highlight specific data based on complex criteria without manual formula creation.

• Prompts to try:

"Highlight the sales just for the enterprise plan, and I want to see which one I have a revenue above 5,000".

"Highlight the revenue cell".

"Color code the age column based on its value. I want to set the smallest value to black and the largest value to white".

3. Creating Columns with New Formulas and KPIs

• Purpose: To quickly introduce new calculated fields or Key Performance Indicators (KPIs) into a dataset.

• Prompts to try:

"Suggest a formula column" (this is an option Copilot provides, often leading to suggestions like "revenue per active user").

"Highlight the churn risk level for each customer segment, including the criteria used, the underlying data signals, and a ranking from low to high risk."

"Could you please go ahead and I want you to calculate the average value of the age column".

"Suggest formulas for this column".

4. Assisting with Complex Formula Construction

• Purpose: To generate sophisticated Excel formulas that might be challenging or time-consuming to write manually.

• Prompts to try:

"Create 12 columns from January to December. And I want in each column to show the monthly revenue of the month".

"Model how a 2 percent increase in our sales growth rate affects gross margin over the next three years.".

"Create a three-year income statement using assumptions sheet GrowthRate, COGSMargin and SG&A as a percent of revenue.".

"Link net income from Year 1 to retained earnings and compute cash flow from operations using changes in working capital.".

5. Building Mini Financial Models

• Purpose: To rapidly construct foundational financial models from a blank sheet, such as headcount or cost projections.

• Prompts to try:

"Help me build a two-year financial model for headcount planning, including salary costs, employer taxes, bonuses, and any other related expenses. Provide both annual and monthly breakdowns, and make sure the model is easy to update for changes in headcount or compensation assumptions."

"Build best-case, base-case and worst-case scenarios with growth assumptions of +5 percent, 0 percent and –5 percent.".

"Run a Monte Carlo simulation with 1,000 iterations for revenue using a normal distribution (mean = €50 million, standard deviation = €5 million).".

"Train a linear regression on the past five years of sales data to forecast next year’s revenue.".

1. https://offers.netgainit.com/hubfs/Microsoft%20Copilot%20Resources/Copilot%20Scenarios%20for%20Finance.pdf?hsLang=en