Prompt Chain: Build a Denial Appeal Workflow From Denial to Submitted Letter

What This Builds

You'll build a 4-step prompt chain — a sequence of AI prompts where each step builds on the previous one — that takes you from a raw prior authorization denial to a complete, ready-to-submit appeal letter in under 10 minutes. Each prompt is saved in a Google Doc so you (or any rep on your team) can run the chain consistently every time. No starting from scratch, no missing steps.

Prerequisites

• Comfortable using ChatGPT or Claude for basic letter drafting (Level 3)
• A Google account for saving the prompt chain as a document
• The denial notice for a current PA you need to appeal
• No paid subscription required — works with free ChatGPT or Claude
• Cost: Free

The Concept

A prompt chain is like a recipe with steps. Instead of one big, complicated AI request that often produces generic output, you break the work into 4 focused steps:

1. Analyze the denial (what exactly is the payer saying?)
2. Research the counterargument (what are the strongest clinical necessity arguments?)
3. Draft the appeal letter (write the formal letter)
4. Review the letter (check for missing elements before submitting)

Each step takes 60–90 seconds. The output of each step becomes the input for the next. The result is a much stronger letter than any single-prompt approach produces.

Build It Step by Step

Part 1: Create Your Prompt Chain Document

Open a new Google Doc and name it "PA Appeal Prompt Chain — [Your Name/Team]."

Create 4 sections with headers: Step 1 | Step 2 | Step 3 | Step 4. You'll paste the prompts below into each section.

Part 2: Write and Save Each Prompt

Step 1: Analyze the Denial

Copy this prompt into your Google Doc under "Step 1":

I'm a patient access representative. A prior authorization was denied and I need to appeal it. Please analyze this denial and tell me:
1. What specific reason is the payer citing?
2. Is this a clinical necessity denial, a coverage/benefit denial, or a procedural denial?
3. What information would most directly counter this denial reason?

Denial details:
- Procedure: [name and CPT code]
- Payer: [insurance company and plan type]
- Denial reason (exact language): [paste exact denial text]
- Patient situation: [2-3 sentences — general clinical context, no full name]

Step 2: Build the Clinical Argument

Copy this prompt into your Google Doc under "Step 2":

Based on the denial analysis above, help me build the strongest clinical necessity argument for this appeal. Please give me:
1. The primary argument (why this procedure is medically necessary despite the denial reason)
2. The supporting evidence points I should reference (what clinical documentation would strengthen this)
3. Any payer policy language or regulatory guidance (e.g., CMS guidelines) that supports covering this procedure

Use the denial analysis from my previous message as the foundation.

Step 3: Draft the Appeal Letter

Copy this prompt into your Google Doc under "Step 3":

Using the denial analysis and clinical argument from our conversation, draft a complete formal prior authorization appeal letter. Include:
- Formal letter header (placeholders for date, payer address, reference numbers)
- RE: line with procedure name, CPT code, and member ID placeholder
- Opening paragraph stating this is a formal appeal
- Clinical necessity argument paragraph (use the points from Step 2)
- Closing paragraph requesting approval and listing supporting documentation being included
- Professional closing

Format it ready to print and submit.

Step 4: Quality Check Before Submitting

Copy this prompt into your Google Doc under "Step 4":

Please review the appeal letter you just drafted and confirm:
1. Does it directly address the specific denial reason? (not just generic "medically necessary" language)
2. Are there any missing elements that payers commonly require in appeals?
3. Is there anything in the letter that could weaken the appeal or give the payer a reason to deny again?
4. What supporting documentation should I attach (list 3-5 most important items)?

Suggest any improvements before I finalize.

Part 3: Test the Chain on a Real Denial

1. Open ChatGPT or Claude in a new conversation
2. Paste Step 1 from your doc, fill in the bracketed fields with a real denial
3. Send and read the analysis
4. Without starting a new chat, paste Step 2 — Claude/ChatGPT will build on what it just analyzed
5. After Step 2's response, paste Step 3
6. After getting the letter draft, paste Step 4 for review
7. Take the reviewed draft and finalize in your word processor

What you should see after all 4 steps: A letter that directly addresses the specific denial reason (not boilerplate), cites appropriate clinical evidence, and has been reviewed for missing elements — in under 10 minutes total.

Real Example: The Full Chain in Action

Starting situation: A prior auth for a CT scan of the abdomen and pelvis (CPT 74178) was denied by Aetna with the reason "Imaging not medically necessary — patient does not meet clinical criteria per InterQual guidelines."

Step 1 — Send and get analysis: Input: [Step 1 prompt with Aetna, CT abdomen/pelvis, InterQual criteria denial, patient with suspected appendicitis symptoms] Output: "This is a clinical necessity denial citing InterQual criteria. The payer is saying the clinical documentation doesn't demonstrate sufficient justification for imaging. The strongest counter would be: (1) specific symptom criteria that meet InterQual imaging criteria, (2) failed or inappropriate alternative workup, (3) urgency of diagnosis."

Step 2 — Build the argument: Input: [Step 2 prompt] Output: "Primary argument: acute abdominal symptoms with fever, elevated WBC, and right lower quadrant tenderness meet InterQual criteria for CT of abdomen/pelvis. Supporting evidence: laboratory values, physical exam findings, clinical notes documenting symptom duration and severity. Regulatory support: CMS guidance that emergency/urgent imaging for suspected acute abdomen does not require prior authorization in most cases."

Step 3 — Draft the letter: Input: [Step 3 prompt] Output: A full, formal 3-paragraph appeal letter addressing the InterQual criteria denial directly, citing the specific clinical indicators that meet the criteria, and noting CMS emergency imaging guidance.

Step 4 — Quality check: Input: [Step 4 prompt] Output: "The letter addresses the denial reason directly — good. Missing: (1) explicit statement of the InterQual criteria being met, (2) mention that clinical documentation is attached. Attach: physician order, lab results showing elevated WBC, clinical notes with physical exam findings."

Time saved: What would have been a 45–60 minute writing and research task took approximately 8 minutes.

What to Do When It Breaks

• Step 2 produces generic arguments (not specific to the denial) → Make sure Step 1 is in the same conversation — the chain only works if all steps are in one continuous chat session; don't start a new chat between steps
• Step 3 letter doesn't reflect the clinical argument from Step 2 → After Step 2, paste a summary: "Using the above analysis, now write the letter" to reinforce the connection
• Step 4 feedback is too vague → Ask specifically: "Is the InterQual criteria directly cited in the letter? Is there any language that could give Aetna a reason to deny again?"
• Letter is too long → Add to Step 3 prompt: "Keep it to 3 paragraphs — concise and direct"

Variations

• Simpler version: Skip steps 1–2 and go directly to Step 3 with a detailed denial description — produces a good letter in one step, though less tailored to the specific denial logic
• Extended version: Add a Step 5 after submission: "Draft a follow-up escalation letter if this appeal is denied again, with a request for peer-to-peer review between our physician and the payer's medical director"
• Team version: Save the completed chain in a shared Google Doc; any rep can run it in their own ChatGPT/Claude session — consistent output quality across the whole team

What to Do Next

• This week: Run the 4-step chain on your next 3 PA denials and compare the quality to your previous appeal letters
• This month: Track appeal success rates — chains that consistently work for specific payers/denial types can be refined and saved as specialized versions
• Advanced: Build separate chains for your most common denial types (InterQual, experimental procedures, out-of-network, wrong setting) — each chain starts at Step 1 but has pre-filled context for that denial category

Advanced guide for Patient Access Representative professionals. These techniques use more sophisticated AI features that may require paid subscriptions.