Faster at the Old Way (Paving the Cow Path With AI)
The Problem
Data lands in the warehouse. The reporting team builds a report, now faster with AI. Someone exports it to a spreadsheet, faster with AI. They clean it, chart it and paste it into a deck, all faster with AI. A vice president prints the deck and reads it with a ruler. Every step a little quicker. The same steps. The reporting team can now build a new report a day, and usage on them is about what you'd expect. AI made the report faster to build. It didn't make it any faster to read. That's paving the cow path: smooth asphalt on a trail that still goes exactly where the cows went.
The Principle
Stop asking for the deck. Ask for the answer, and what the person would do about it. Using AI to build more reports faster gets you micro little efficiencies down the same old path. Throw it at the whole chain instead of one step at a time. Find the real question behind the report, and answer that question directly, with its sources, in seconds. If you lead people, this change costs nothing: stop asking for the PowerPoint or the Excel file. If you're the one building the same deck every week, make something different once and see what people think. Better yet, make it interactive, a page people can click through and ask questions of. Nobody ever did a single thing differently because the slide had nicer icons.
4 Steps to Take Action
- 1
Find the VP with the ruler
Sit with someone while they read the report you send them. Ask what question they're actually trying to answer.
- 2
Time the answer
From the moment they ask to the moment they have the answer in hand. Write it down.
- 3
Answer the question directly
One question, one leader, on data you trust, with every answer citing its sources.
- 4
Count the reports AI built
And how often each one is opened. Retire the ones nobody reads, the way you'd kill a dead dashboard.
💡 Inspired By
Chapters 2, 6 and 8 of The Second Factory. And a vice president I once watched read a metrics report with a wooden ruler and a yellow highlighter, one line at a time, hunting for the answer to a question the report had never asked him.
This is what Analytics Pitfalls & Principles is about: people helping people.
In Practice
In her first weeks leading AI at ClassicMotors, Nora Berg kept meeting people who were thrilled that AI could build their PowerPoints. A sales manager cut his quarterly business review deck from two days to an afternoon. A product lead made twelve slides with matching icons from a spreadsheet in nine minutes. Nobody could tell her one thing anyone had done differently because of them. In her notebook she wrote: Faster at the old way. A young analyst on the reporting team showed her a new weekly report AI had built in an afternoon. "I've got three more in the queue. I can build one a day now." A few months later, the data team counted sixty-one new reports built with AI. Usage on them: about what you'd expect. Then she sat in the VP of Marketing's office while he read one of them. Eleven pages, forty-one columns. He printed it, set a wooden ruler under the first row and slid it down the page with a yellow highlighter in his other hand. It took him most of every Tuesday morning. He was looking for one thing: past sellers who were about to list again and hadn't heard from him.
