Results depend on the tool, the model and what you enter.
Result preview
Example: written by our editors, not real AI output.
Check: fraud alert
Real cases: 50 of 10,000
Found: 45
False alarms: 500
Real share of hits: 8 %
Next: second check before blocking
Fill in
What you type stays in this browser window. We store and send nothing.
What is checked?
fraud alert
lead score
quality test
churn warning
mixed
How rare is the case?
under 1 %
1 to 5 %
5 to 20 %
unclear
mixed
What follows a hit?
manual review
block
offer
escalation
mixed
Your prompt
Help me read a hit rate with the base rate in mind.
Check: [Check]
Base rate: [Base rate]
Decision: [Decision]
Steps: 1. State how often the thing you look for really occurs in the checked group. 2. Write down how often the check finds real cases and how often it raises a false alarm. 3. Convert everything into counts for 10,000 cases. 4. Calculate how many positive results are real and how many are false. 5. Check whether the base rate differs between groups or over time. 6. Recommend what should happen after a positive result, such as a second check.
Output: count table, share of real hits, effect on the decision and next step.
Take particular care to turn rates into counts before judging a positive result.
Before you answer, check: Are all instructions met? If key information is missing or a placeholder in square brackets is still in the text, ask instead of guessing. Mark assumptions as assumptions.
You have edited the prompt. Changes in the fields above no longer overwrite it.
This is an instruction for an AI, not the result itself. Paste the prompt into a chat in ChatGPT, Claude or Gemini and send it.
Prompt that finds label cardinality in metrics and proposes cuts plus a budget.
Prompt · Intermediate · about 15 min
You may use our prompts freely, including at work. Please do not republish them as your own collection. We give no warranty for AI results; check them yourself. Terms of use