Gen AI Tips: How to Use Microsoft Copilot With Microsoft Excel
How to Supercharge Excel Data Analysis with Microsoft Copilot, Generative AI, 1.5×IQR Statistical Outlier Rule
Microsoft Excel has always been a powerful tool for data analysis — but with Microsoft Copilot and Generative AI, you can now automate deeper insights, streamline repetitive tasks, and uncover patterns that would have taken hours to find manually. This combination of Excel and Copilot was central to my 311 Toronto Customer‑Initiated Service Requests analysis — a project I document on my civic analytics blog, We Protect Toronto. By applying the 1.5×IQR Statistical Outlier Rule, which is a core focus of this blog, I was able to isolate unusual FSAs and identify meaningful trends in the 2010–2025 time series. Copilot then added the contextual intelligence needed to understand why those outliers were emerging, enriching the analysis and making the insights easier to communicate to City of Toronto councillors and staff.
Whether your Excel file contains formulas, Power Query logic, VBA, or just raw data, Copilot can analyze named ranges, generate code, summarize trends, and even integrate external information such as demographics or media coverage.
Below are five practical, real‑world examples of how Copilot can transform your Excel‑based analytics workflows.
1. FSA‑Level Analysis with Excel and Generative AI
When you’re working with neighbourhood‑level data, Excel is often the fastest way to isolate patterns before bringing in generative AI for deeper context. For example, suppose you’ve analyzed the average dollar value of uncollected parking tickets across Toronto FSAs over the past year. Using the Statistical Outlier Dashboard from the Interactive Statistics Education Add‑In, you identify five FSAs with unusually high uncollected revenue.
One of these might be M4A.
This workflow mirrors the approach I used when analyzing 311 Toronto Customer‑Initiated Service Requests at the FSA level — a project I wrote about on my civic analytics blog, We Protect Toronto. I began in Excel, using descriptive statistics and outlier detection to flag FSAs with unusual service‑request patterns. Once M4A emerged as an outlier, I opened Microsoft Copilot directly inside Excel and asked:
“Provide a demographic analysis of FSA M4A.”
Copilot instantly generated a detailed demographic profile using publicly available data sources such as Statistics Canada Census Profiles, including:
- Age and gender distribution
- Ethnic and cultural diversity
- Language and religion
- Housing and income
- Education and employment
Pairing Excel’s quantitative analysis with Copilot’s contextual insights made it much easier to understand why M4A behaved differently — whether in parking ticket collections or 311 service‑request patterns. This combination of statistical detection and AI‑driven interpretation gives you a richer, more actionable understanding of neighbourhood‑level trends.
2. Call Centre Interaction Analysis with Excel and Generative AI
Call centre datasets often contain fields such as CALL REASON, which can reveal operational patterns when analyzed properly. Excel is excellent for summarizing these interactions, and Copilot can extend that analysis by adding contextual intelligence when needed.
This mirrors part of my workflow during the 311 Toronto Customer‑Initiated Service Requests analysis — a project I document on my civic analytics blog, We Protect Toronto. In that work, I focused on identifying long‑term trends and FSA‑level patterns using Excel and the 1.5×IQR rule. While I did not perform sentiment analysis in that project, Copilot can classify interactions by sentiment when a dataset includes the appropriate fields. This capability would allow analysts to explore whether certain call reasons consistently generate more positive or negative experiences.
For example, if your dataset includes:
- FSA
- CALL REASON
- a sentiment field (if available)
- flags for unusually negative interactions
…you can ask Copilot to summarize interactions by CALL REASON and sentiment, then use the Statistical Outlier Dashboard to identify which call reasons produce statistically unusual patterns.
This workflow helps you pinpoint:
- service issues
- communication gaps
- operational bottlenecks
- policy‑related frustrations
Even if your dataset doesn’t include sentiment, Copilot can still help you categorize call reasons, group similar issues, or generate new variables that make the analysis easier to interpret — just as it did in my 311 Toronto work.
3. Media Monitoring with Excel and Generative AI
Media monitoring is one of the most powerful ways to add context to the patterns you uncover in Excel. Once you’ve identified unusual trends at the FSA level, Copilot can help you understand whether broader events in the city might be influencing what you’re seeing.
This is exactly the approach I used when analyzing 311 Toronto Customer‑Initiated Service Requests at the FSA level — a project I document on my civic analytics blog, We Protect Toronto. After isolating several FSAs with unusual activity in Excel, I wanted to understand whether anything happening in the city during that period might help explain the spikes.
Instead of switching tools, I opened Microsoft Copilot directly inside Excel and asked:
“Search social and mass media for news affecting my organization over the past month.”
Copilot can scan public news coverage and summarize events affecting the City of Toronto, including:
- major incidents or disruptions
- infrastructure issues
- policy announcements
- neighbourhood‑specific events
- citywide trends that may influence service‑request volumes
You can even specify time periods (“over the past month,” “since January,” “during the summer”) or geographic focus (“in Scarborough,” “in North York,” “around FSA M4A”) to align the media scan with the exact window of your Excel analysis.
By pairing Excel’s quantitative trend detection with Copilot’s media‑contextual insights, you gain a clearer understanding of why certain FSAs behave differently. This combined workflow turns raw patterns into actionable, city‑level intelligence.
4. Recoding Variables Automatically with Excel and Generative AI
Large datasets often require recoding — a tedious task even for experienced analysts. Copilot can turn this into a fast, guided workflow that saves hours of manual effort.
This was especially useful during my 2010–2025 time‑series analysis of 311 Toronto Customer‑Initiated Service Requests, which I document on my civic analytics blog, We Protect Toronto. While analyzing the data in Excel, I wanted to enrich the dataset by adding a new variable that assigned each FSA to one of the six former City of Toronto municipalities. Ward information was already included in the original dataset, but I needed a second geographic lens to help councillors and staff understand where service‑request hotspots were emerging within their broader municipal boundaries.
Instead of manually building a lookup table or writing a long nested formula, I asked Copilot directly inside Excel to help me match each FSA to its corresponding Canada Post FSA and LCD Route Reference, which in turn could be linked to a specific municipality. This allowed me to generate a clean MUNICIPALITY variable that made the analysis richer, more interpretable, and easier to communicate to decision‑makers.
5. Financial Statement Analysis with Excel and Generative AI
Excel and Copilot can work together to build powerful ratio‑style scenario analyses — even outside traditional financial statements. This approach was especially useful during my 2010–2025 analysis of 311 Toronto Customer‑Initiated Service Requests, which I document on my civic analytics blog, We Protect Toronto. Because the dataset is available from the City of Toronto Open Data Portal and includes the number of calls assigned to each of the City of Toronto’s ~30 Divisions, it’s straightforward to build a financial‑style model that estimates operational cost trends over time.
In Excel, you can structure the analysis like this:
- Row 1: Number of Customer‑Initiated Service Requests (per Division, per year)
- Row 2: Estimated Cost per Call — generated with Copilot’s help by asking for a reasonable estimate for a publicly funded inbound call‑centre operation
- Row 3: Total Cost per Call — calculated by multiplying Row 1 and Row 2
- Row 4: Cost‑per‑Call Growth Rate — Copilot can calculate this automatically for each Division across the 16‑year time series
Once these rows are in place, you can ask Copilot to analyze the trends:
“Analyze the trend in Total Cost per Call and Cost‑per‑Call growth for each Division.”
Copilot can then:
- identify Divisions with unusually high cost growth
- highlight years where operational costs spiked
- flag anomalies that may correspond to policy changes, service disruptions, or workload surges
- generate a narrative summary explaining the long‑term trend
Final Thoughts
Microsoft Copilot transforms Excel into a far more powerful analytics environment — one where statistical methods, operational data, and contextual intelligence all come together without ever leaving your workbook. Across the examples in this post, the combination of Excel, the 1.5×IQR Statistical Outlier Rule, and Copilot mirrors the workflow I used in my long‑term 311 Toronto trend analysis: isolate patterns with Excel, identify outliers with IQR, and then use Copilot to explain why those patterns might be emerging.
Excel remains the engine — Copilot becomes the accelerator. Together, they allow analysts, educators, and civic decision‑makers to move from raw data to meaningful insight with far greater speed and clarity.
Visit We Protect Toronto for examples →