LEADERSHIP JUDGMENT CHALLENGE 002
THE REPORT
NOBODY HANDED OVER
Six months after an AI drafting tool launched, the adoption dashboard says usage is still below threshold. Only 12% of reports start from an AI-generated draft.
What would you do?
THE SITUATION
The tool tested well. Adoption didn't follow.
Six months ago, your organization rolled out an AI drafting tool for reports.
Testing showed the drafts were accurate, well-structured and needed only light editing. The results were validated before launch.
Today, usage tells a different story. Only 12% of reports start from an AI-generated draft. The other 88% are still built by hand. The adoption dashboard recommends a refresher training — and the budget for it is already approved.
Only 12% of reports start from an AI-generated draft.
MAKE THE CALL
What do you do?
Choose before you continue.
WATCH THE CHALLENGE
Now test your decision.
Hold on to your choice. The challenge isn't whether the usage number is accurate. It's whether you know what it's actually evidence of.
THE JUDGMENT PROBLEM
A metric is not an explanation.
The usage data may be completely accurate. What it's evidence of can still require judgment.
The evidence is real. The cause isn't established. 12% could mean a lack of skill. It could mean a lack of trust. It could mean people found a workaround that works fine for them. Or it could be something the dashboard simply can't see.
Same number. Not the same cause.
The wrong diagnosis can cost more than the problem itself. Training fixes a skills gap. It won't touch whatever else is going on — and right now, you don't yet know which one this is.
THE PRESCRIPTION TRAP
Don't treat the recommendation as the diagnosis.
If more than one answer fits the number, you don't have a diagnosis yet.
The dashboard can measure usage perfectly and still be wrong about what's causing it. A recommendation attached to a metric is a guess wearing the metric's authority — not a finding.
THE JUDGMENT DIFFERENCE
"Is it low" is not the same question as "why."
Is usage low?
Why is usage low?
This isn't about whether people can learn the tool. It's about what's actually stopping them.
BETTER QUESTIONS
Before you act on the recommendation, ask:
What is this metric actually measuring?
What else could produce this exact same number?
If the fix didn't move the number, what would that tell me?
If you can't rule out at least one competing explanation before you act, you're prescribing from the symptom — not the cause.
SO, WHAT WOULD I DO?
Investigate what's actually driving the low usage.
Not because training is the wrong idea, and not because six months is definitely enough time. I don't yet know which explanation fits.
Investigate doesn't mean delay. It means getting the information every other option is guessing at, before I commit budget and change management to a fix that might not touch the real cause.
THE LEADERSHIP JUDGMENT TAKEAWAY
Don't prescribe from the symptom.
Rule out competing explanations before you act.
Treat the recommendation as a guess, not a diagnosis.
Accurate evidence can still lead to the wrong action.
If you misunderstand what it represents.
BRING THE CHALLENGE TO YOUR TEAM
Make judgment visible.
Judgment Challenges can be explored with your team through a private 90-minute Judgment Lab, turning the scenario into a practical conversation about metrics, diagnosis and the fixes we reach for too quickly.