When Copilot Can’t Copilot: A Field Report on Excel AI’s Limits

Microsoft 365 Copilot in Excel is positioned as a transformative productivity tool.  For association finance teams, the promise is clear: describe what you need in plain language, and Copilot will analyze your data, generate charts, and even explain the results.  The value proposition is especially strong for reporting workflows, where visuals like budget comparisons and trend analyses are commonplace and impactful.But promises only matter when they hold up under real conditions.  This article was originally intended to highlight a few basic visualization use cases for Excel Copilot in association finance.  Instead, it became something more useful: a firsthand account of what happens when Copilot does not deliver, and why good judgment, not AI, is still a finance team’s most important asset.  One specific example illustrates the point.

Just a Basic Chart

The objective was straightforward.  Using a structured Excel table named BudgetVsActuals, the goal was to generate a clustered column chart comparing budget and actual expenses by month for 2025.  The data was clean, clearly labeled, and formatted for easy analysis.  The initial prompt was specific and comprehensive:

Using the table named BudgetVsActuals, create a clustered column chart that compares Budget and Actual expenses by month for 2025. Format the chart for presentation use: add a descriptive title, apply professional color formatting with distinct colors for each series, include data labels on each column, and format the Y-axis as currency with no decimal places. Set the X-axis to display all months in order from January to December, and include a legend positioned at the bottom. Title the chart “2025 Monthly Expenses: Budget vs Actual.”

What followed was not a chart.  Copilot responded:

I encountered an issue while trying to create the chart. Please ensure the table "BudgetVsActuals" is correctly formatted and try again.

The same message appeared after rewording, simplifying, and reducing the prompt.  A final attempt asking Copilot to explain the error resulted in:

I am unable to directly create the clustered column chart because I cannot execute actions like generating charts or applying formatting within Excel. However, I can guide you step-by-step on how to create the chart using the "BudgetVsActuals" table in your worksheet.

At this point, the assistant had abandoned the task altogether.

Even with a clearly structured table and detailed prompt, Copilot failed to produce a simple chart and offered no actionable explanation.

A Pattern, Not an Exception

The issue was not a malformed table, a poorly written prompt, or a complex task.  While Copilot has successfully produced visuals in other cases, the output often lacked polish or required cleanup to be usable.  In most cases, the chart could have been built faster and with better formatting by hand.

That is the crux of the problem.  Even when Copilot works, it often does not add value.  And when it fails, it wastes time.  There is no meaningful error messaging, no suggested alternatives, and no clear indication of whether the issue is with the data, the request, or the tool itself.

This level of inconsistency makes Copilot unreliable for core visualization tasks. Finance teams operate under time constraints and with high expectations for clarity and accuracy.  If a tool routinely fails to complete basic tasks or produces substandard output, it cannot yet be trusted for professional workflows.

The Bigger Lesson

This is not a rejection of Excel Copilot or AI.  As demonstrated in our previous article, Excel Copilot for Associations: Getting Started, Copilot performed well with data manipulation, such as generating advanced formulas, applying conditional formatting, and even referencing lookup data.  In that respect, it is certainly a time saver, especially for staff members with basic Excel proficiency who want to leverage advanced functions.

The future of Excel will undoubtedly include more automation, smarter assistants, and more sophisticated visualization tools.  Progress is coming quickly.

But today, teams must be clear-eyed.  It is important to know when AI is actually saving time, and when it is simply absorbing it.  Sometimes, reaching for AI because it is available results in a slower, more frustrating experience than doing the work manually.

Closing Notes

While Excel Copilot shines in certain areas, especially with formulas, summaries, and structured data manipulation, it is not yet ready to serve as a dependable assistant for financial charting.  It will likely get there soon, and when it does, SoundPost will be ready to share its strengths with the association community.

In the meantime, association finance teams should continue moving full steam ahead on AI adoption, including proven use cases of Excel Copilot.  Although not every tool is ready for every task, the opportunities for efficiency, insight, and innovation are real and growing quickly.