Copilot in Operations: how AI changes the pattern

For years, reporting worked the same way on most teams I’ve been part of. A business user — someone who owns a process but doesn’t write formulas — explains the logic to an analyst. The analyst builds it in Excel. Then they go back to the business owner to review the output and collect feedback. A few rounds of that, and you have a report.

It’s a pattern that worked for a long time. It also turned a handful of people into Excel stars — the ones who knew exactly how a workbook was wired together. Genuinely useful people. Right up until the workbook breaks and you realise they’re the only ones who understand how it was built.

What changed when Copilot arrived

When Copilot landed in Excel, the analytics team didn’t disappear. We still did the heavy lifting — the parts that are actually hard. Going out to several source systems, pulling the data, and stitching it into one workbook. That’s where the real value sits, and it’s where a person still adds something.

What changed is the cost of the work around it. A data extraction that used to take 10–15 minutes — a long sequence of clicks, each one a chance to pick the wrong filter or paste into the wrong cell — now takes up to 5 minutes, with room to improve further. Accuracy went to 100%: no copy-paste errors, and Excel now exports directly to CSV, which feeds straight into the downstream system that consumes it. Fewer clicks, fewer places to make a mistake, and a process you can trust to run the same way twice.

The twist: talking to Copilot, not to me

Here’s the part I didn’t expect. Working with my colleague Anastasija Kondratjeva, we did the heavy lifting together — connecting the sources and shaping the base of the workbook. Then, instead of having her explain the business logic to me so I could translate it into formulas, I showed her how to explain it directly to Copilot.

Copilot did the complex formula writing. And when something stopped working, Anastasija could fix it herself and keep the daily operation running — without raising a ticket and waiting on the Data and Analytics team to free up. The dependency that used to define the whole pattern just quietly went away.

Why this matters

This is what people mean when they say efficiency is becoming a commodity. The specialist skill — translating business logic into working formulas — is no longer the bottleneck. It’s available to whoever owns the process. That’s not a threat to the analytics team; it frees us for the work that’s actually hard, and it lets smart people in every department drive their own improvements.

The saved clicks aren’t the point. The point is what they free up, and who they free up. When a team can run its own operation faster and without errors, the downstream effects are real: fewer delays, more reliable data, and a process that doesn’t stall when one person is unavailable. That shows up where it counts — in a better experience for the customer.