Draft a lead-scoring rubric as a JSON config

About this exercise

Marketing Ops will paste your lead-scoring draft into their tool, so you need it as one JSON code block: an array of objects with signal, points and reason fields. Use only the signals your analysis found, points must add up to exactly 100, and leave out company headcount because the purchased list it came from is unreliable. After the code block, write one sentence under 25 words asking Sales leadership to review the weights. At most 12 lines of code.

Common mistakes when you ask AI for this

• Not naming the source signals. The AI adds popular scoring signals like webinars and email opens because most models use them. Try: Add: "Use only the signals the analysis found predict a win."
• Missing the headcount problem. The AI scores headcount because it predicts wins in the data. It can't know the list is unreliable. Try: Add: "Leave out company headcount; the purchased list is unreliable."
• No points total. The AI picks round numbers that add up to 85 or 120. Try: Add: "Points must add up to exactly 100. Check the sum."
• Explaining around the code. The AI writes paragraphs before and after the JSON. Try: Add: "After the code block, one sentence under 25 words. Nothing before it."
• Long JSON. The AI pretty-prints with every brace on its own line, so the block runs long. Try: Add: "One object per line, at most 12 lines."

What a strong prompt includes

• Asks for JSON
• Sets the fields
• Sets the total
• Drops headcount
• Limits the prose

Practice this now ↑

Related exercises

• Accept the buyer's counteroffer on three stated conditions (Sales AI · Sales and customer success)
• Accrue goods received but not invoiced at PO price (Finance AI · Accountants)
• Add dark mode by overriding your existing CSS variables (Coding AI · Software developers)
• Add input checks to your number guessing game (Coding AI · High school students)
• Add labels to a signup form without rewriting the page (Coding AI · High school students)