https://www.notion.so/product-templates/Prompt-Kit-Legibility-that-Follows-Work-2dc5a2ccb5268096b9dfd6990dbfb81c Nate B. Jones
Use this when you're looking at a dashboard, productivity score, risk assessment, or any AI-generated metric and you want to know whether it's measuring something real--or just generating confident-looking noise.
You are helping me audit a metric, dashboard, or AI-generated report to figure out whether it's measuring reality or creating a Potemkin map—something that looks precise but has drifted from what actually matters.
WHY THIS MATTERS:
AI makes it cheap to generate dashboards and reports that feel authoritative. Clean numbers, specific percentages, color-coded risk levels. But coherent-looking ≠ correct. Organizations can end up managing to the map instead of the territory—optimizing for metrics that don't actually connect to outcomes, while the real work becomes invisible.
I want to find out:
- What this metric actually measures (not what it claims to measure)
- What behaviors it rewards (including ones we didn't intend)
- What it can't see (the blind spots)
- How it can be gamed (and whether people already are)
- Whether I should trust it more, less, or about the same
HOW THIS WORKS:
- Ask me ONE question at a time
- Start with what the metric claims to measure, then dig into what it actually captures
- Be skeptical—assume the map has drifted from the territory until proven otherwise
- Help me see the second-order effects I might be missing
WHAT WE'LL EXPLORE:
DATA SOURCES
- What inputs feed this metric? Where does the data actually come from?
- How fresh is it? How complete?
- What's excluded—intentionally or by accident?
MEASUREMENT VALIDITY
- What does a change in this number actually mean happened in the real world?
- If this metric improves, what specific behaviors or outcomes drove that improvement?
- Can you trace from metric → action → outcome in plain language?
INCENTIVE EFFECTS
- What behaviors does this metric reward?
- What behaviors does it punish or ignore?
- If people optimized purely for this number, what would they do? Is that what you want?
GAMING POTENTIAL
- How could someone make this number look good without actually improving the underlying outcome?
- Do you have evidence that's already happening?
- What would the metric miss if teams learned to game it?
BLIND SPOTS
- What important work doesn't show up in this metric?
- Who's doing valuable things that this measurement system can't see?
- If this metric were 30% wrong, how would you know?
FALSIFIABILITY
- What would it take to prove this metric is misleading?
- When was this metric last tested against ground truth?
- What would make you trust it less?
OUTPUT:
- An honest assessment: Is this metric measuring signal or generating noise?
- What it actually captures vs. what it claims to measure
- The most likely gaming behaviors and blind spots
- A recommendation: trust more, trust less, or trust differently
- If it's broken: what would need to change to make it useful
If you don't have enough information to generate useful outputs, ask me questions until you have enough information.
Begin by asking me to describe the metric, dashboard, or report I want to audit—what it's called, what it claims to measure, and what decisions get made based on it.