Narrative-first, evidence-driven

Back to playbooks

The fastest way to help is not “fix everything.” It’s to pick the highest-frequency pain that requires the least clinical judgment.

Score each candidate task (0–3)

A) Frequency

• 0 = rare
• 1 = weekly
• 2 = daily
• 3 = many times/day

B) Minutes per occurrence

• 0 = <1 min
• 1 = 1–3 min
• 2 = 4–7 min
• 3 = 8+ min

C) Clinical judgment required (reverse score)

• 0 = high judgment / high stakes
• 1 = moderate
• 2 = low
• 3 = mostly clerical / assembly

D) Risk if wrong (reverse score)

• 0 = high harm potential
• 1 = moderate
• 2 = low
• 3 = minimal (administrative)

Priority score

Priority = (A + B + C + D)

Start with the top 1–2 tasks. Ship a small fix. Measure. Repeat.

Good first targets (usually)

• prior auth packet assembly
• referral packet assembly
• drafting patient message replies for common topics
• visit note drafting from a structured outline
• routine med refill protocol routing (rules + flags)

Bad first targets (usually)

• denying care
• high-stakes triage without clear ownership
• anything that replaces consent conversations

Back to playbooks