Score a backlog with RICE using AI (without inventing the numbers)

First

Paste your candidate features (one per line) after the prompt, into ChatPRD or any strong LLM.

You are helping me prioritize a product backlog with the RICE framework
(Score = Reach × Impact × Confidence ÷ Effort).

Rules:
- Do NOT invent Reach or Impact numbers. For each item, ask me for the
  data you need (e.g. "how many users/quarter does this touch?").
- Use Impact on the standard RICE scale (3 = massive, 2 = high, 1 = medium,
  0.5 = low, 0.25 = minimal).
- Confidence is a percentage (100/80/50). Effort is person-months.
- Mark any value I don't give you and you had to assume with [ASSUMPTION].

Output a markdown table sorted by RICE score descending, plus a one-line
rationale per row. After the table, list the 3 items whose ranking is most
sensitive to an assumption I should go verify.

Here is the backlog:
[paste your features here]

Then

Pull the real Reach numbers for the 3 sensitive items from analytics (not memory), re-run, then gut-check the top of the list — if something strategic ranks low, override it with a written reason.