Earlier this year I compared Google Colab and Excel Copilot for analyzing Boston Crime Statistics (Google Colab vs Excel Copilot). This time I tried the same data set with Excel Labs Agent Mode and it was a completely different experience in Excel.
With the same dataset – 260,000 records of Boston crime incidents – and the difference is night and day. Where Copilot stumbled and failed, while Agent Mode delivered a complete analysis with explanations and recommendations, all while staying comfortably within Excel.
What is it?
Excel Agent Mode is Microsoft’s latest addition to Excel Labs (part of the Frontier program), and represents a significant evolution in conversational data analytics within spreadsheets. Think of it as having a data analyst embedded directly in your Excel environment who actually understands what you’re asking.
Agent Mode operates with a more sophisticated reasoning engine. It can handle complex, multi-step analytical tasks through natural language prompts. You simply ask your question, in my case “Please summarize these statistics and help me understand and predict future occurrences” – and Agent Mode develops an analysis plan, executes it, handles data quality issues on the fly, and delivers comprehensive insights. Use with care though as it is operating on your Excel spreadsheet.
The tool works entirely within the familiar Excel interface, requiring no Python coding or external notebook environments (using the M365 Analyst Agent requires stepping outside of Excel – I repeated this experiment in the Analyst Agent and it also did a great job). It’s designed specifically for business users who need powerful analysis capabilities without leaving their spreadsheet comfort zone. From my testing with the Boston Crime Statistics dataset, Agent Mode successfully processed all 260,000 records, identified patterns across neighborhoods and time periods, and even provided actionable recommendations, something that Copilot in Excel couldn’t quite finish.
What does it mean from a business perspective?
The implications of this technology for organizations are substantial:
- Democratized Advanced Analytics: Business analysts, managers, and other professionals who work primarily in Excel can now perform more sophisticated data analysis without requiring Python skills or data science expertise (up to a point). This dramatically expands who can extract meaningful insights from complex datasets.
- Increased Productivity at Scale: Agent Mode handles reasonable datasets (hundreds of thousands of rows) which means reduced time-to-insight for larger sets of operational, sales, financial, and customer data.
- Reduced Tool Fragmentation: Organizations can keep analytical work within the Excel where a lot of business data already lives, eliminating the productivity drain of constantly switching between tools or waiting for technical teams to run analyses.
- Built-in Error Handling and Data Quality Management: The Agent Mode actively seemed to identify and addresses data inconsistencies and problems during analysis, reducing the risk of decisions based on flawed data – a significant advantage over manual approaches.
- Strategic Decision Support: By providing not just descriptive statistics but also pattern recognition and predictive insights (as demonstrated with the temporal and geographic crime patterns – the Analyst Agent scored here by automatically graphing the data), Agent Mode enables more forward-looking business planning.
- Cost Efficiency: Organizations may reduce their dependence on specialized BI tools or external data science resources for routine analytical needs, optimizing their technology spend.
What do I do with it?
Here’s an action plan to get started:
- Enable Excel Labs: Excel Agent Mode is currently available through Excel Labs (the experimental features add-in from Microsoft) in the Frontier program. Install Excel Labs from the Excel Add-ins menu – it’s free with your Microsoft 365 subscription.
- Start with a Pilot Project: Identify a current analysis challenge in your organization – perhaps a sales trend analysis, operational metrics review, or customer behavior dataset. Choose something meaningful but manageable to test Agent Mode’s capabilities – even something you can validate against previous analysis.
- Develop Clear Natural Language Prompts: Be specific about what you want to understand. Instead of “analyze this data,” try “identify the top 10 customers by revenue, show seasonal trends, and flag any unusual patterns.” The more context you provide, the better the results.
- Review the Analysis Plan: When Agent Mode responds to your prompt, it shows you its proposed approach before executing. Take a moment to review this – it’s a great learning opportunity and ensures the analysis aligns with your intent. As the Agent is working also expand the steps to take a look at what it is doing.
- Iterate and Refine: Don’t expect perfection on the first try. Like my experience with the Boston data, you may need to adjust your approach or provide additional context.
- Share Results Responsibly: While Agent Mode handles the technical heavy lifting, you still need to apply business judgment to the insights. Validate recommendations against your domain knowledge before acting on them.
My less than stellar previous experience with Excel Copilot and this particular dataset has been redeemed by Agent Mode. What couldn’t be accomplished with Copilot – analyzing 260,000 crime records, identifying patterns, and generating actionable recommendations – was completed smoothly with Agent Mode, all without leaving Excel.
For business professionals drowning in data and needing an extra hand, Excel Agent Mode brings the power of conversational data analytics to your familiar spreadsheet environment, and based on my testing, it actually works.