The Spreadsheet AI Trap: Why AI-Powered Excel Isn't Enough
Microsoft Copilot in Excel sounds perfect for retail. In practice, it solves the wrong problem.
Ersel Gökmen
December 26, 2025
Microsoft announced Copilot in Excel and every retail analyst cheered. Finally, AI that works with the tool they already use! "Create a pivot table showing revenue by store." Natural language Excel operations.
It's genuinely useful. And it's genuinely insufficient.
The Single-File Problem
Excel Copilot works within one file. Retail analysis rarely lives in one file. "Compare Shopify sales to the purchase orders in the ERP export against the inventory file from the warehouse" — that's 3 files, 3 formats, 3 schemas. Copilot can't join them.
The Analysis-to-Action Gap
Even when Copilot produces a great analysis in Excel, then what? You still need to copy numbers into an email, format a report in PowerPoint, update prices in Shopify, and send a Slack message. Excel Copilot's scope ends at the spreadsheet border.
The Right Layer
AI assistance at the spreadsheet level is a local optimization. The global optimization is AI at the workflow level — connecting data, analysis, and action in one system. That's the layer where retail value is created.