What it does

Copilot in Excel analyses data, suggests formulas, adds formula columns, highlights and sorts, and generates charts and PivotTables from a natural-language prompt. There's also a Python-in-Excel path for heavier analysis.

Key facts

• It works best against data formatted as an Excel table. Loose ranges give patchy results.
• It can write and explain formulas, and drop a new calculated column in for you.
• Advocated analysis (trends, outliers, breakdowns) returns charts and PivotTables you can keep.
• Copilot with Python can run more serious statistical work in the cloud, licence permitting.

When to use / skip

Good for exploring an unfamiliar dataset, drafting a formula you can't quite remember, and quick visualisation. Skip it as an oracle for numbers you must trust blind — always sanity-check the maths.

Configuration decisions

• Whether Python-in-Excel is allowed in your tenant.
• Guidance on formatting-as-table so people actually get value.

Gotchas

• No formatted table, no joy — this is the single biggest reason it "doesn't work" for people.
• It can misread column intent and produce a plausible but wrong formula. Check the logic, not just the result.
• Very large or messy workbooks blunt it noticeably.

Consultant notes

• The table-formatting prerequisite trips up nearly everyone. Lead your training with it.
• Finance teams are rightly wary of trusting generated calculations — position Copilot as a drafting and exploration aid, with human verification non-negotiable.
• Python-in-Excel is powerful but a governance conversation of its own; don't switch it on without thinking about data egress.

---

Worth revisiting after the next release wave.