Microsoft's Own Copilot Data Can't Answer the Question Microsoft Isn't Asking

Work Trend Index reports keep measuring how much faster Copilot users draft documents. None of them ask whether the documents needed to exist, and economists already have a name for what happens next.

Microsoft's Work Trend Index measures drafting speed, not whether the recovered time bought better decisions.

A manager watches Copilot cut a quarterly report's drafting time from a day to an hour and mentions it proudly at the next all-hands. Nobody in the room asks the harder question: whether the report needed a reader in the first place, or whether the organization is now just producing the same volume of mediocre reporting three times faster than before.

Productivity is a measure of output per unit of input. It says nothing about whether the output was worth producing. Microsoft's annual Work Trend Index reports track exactly this category of gain — drafting speed, summarization time, meeting recap turnaround — and stop there, because output-per-hour is the number that's easy to measure and speed is the story that's easy to tell.

Economists have a name for what tends to happen when a resource gets cheaper to use: the Jevons paradox, first observed with coal efficiency in the nineteenth century, where making something more efficient increases total consumption of it rather than reducing it. Applied here: cheaper drafting doesn't produce less total drafting. It produces more documents, more reports, more of whatever the tool made faster, because the freed capacity gets absorbed into more volume rather than banked as time.

Individual workers experience that absorption directly. Time saved by AI-assisted drafting rarely becomes deep, unhurried thinking; it becomes capacity that a manager or a system reassigns to more output, because throughput is what gets measured and judgment is not.

Some organizations are making the harder, deliberate choice instead — redirecting freed time toward fewer, better client reviews rather than more client reports, using recovered capacity for the judgment work that speed alone can't produce. That requires an explicit decision by someone with the authority to make it, because left to inertia, the default outcome is simply more of the same output, produced faster.

The honest test of a Copilot rollout isn't the chart in the next Work Trend Index report. It's whether headcount, scope, or output volume quietly grew to absorb the saved hours within a year of deployment. Most organizations don't publish that number, and the reason is probably that it would answer the question the productivity report was designed not to ask.