Prompt Chain: Automate Personalized Quarterly Report Commentary

What This Builds

A repeatable workflow that lets you generate personalized quarterly commentary for 50-150 client accounts in under 2 hours — instead of spending a full day (or skipping it entirely). Instead of writing one letter at a time, you create a template prompt, prepare your data in a spreadsheet, and batch-generate personalized commentary for every client using a consistent process.

Prerequisites

• Comfortable with basic ChatGPT prompting (Level 1)
• ChatGPT account (free works; Plus gives faster response for batch work)
• Microsoft Excel or Google Sheets for your data table
• Your portfolio management system (Orion, Black Diamond, etc.) for performance data export

The Concept

A prompt chain is like an assembly line for writing. Instead of writing each client's letter from scratch (slow, inconsistent), you build a template once, fill in the client-specific variables, and run it through ChatGPT for each client.

Think of it like mail merge — you probably know how to use Word's mail merge to send personalized emails. This is the same idea, but instead of just inserting names and addresses, you're inserting portfolio performance, life events, and financial planning context — and getting a full personalized paragraph instead of just a filled-in blank.

Build It Step by Step

Part 1: Build your data table

Open Google Sheets and create a table with one row per client. Minimum columns needed:

Populate this table from your CRM and your portfolio management system. For your first run, do 10 clients to test the workflow.

Part 2: Write your master template prompt

In a text file or Google Doc, write your base prompt. This is the template you'll fill in for each client:

Write a personalized 150-word quarterly performance commentary for a wealth management client.

Client: [CLIENT FIRST NAMES]
Portfolio performance this quarter: [PORTFOLIO RETURN]
Benchmark comparison: [BENCHMARK RETURN]
Portfolio activity: [NOTABLE PORTFOLIO CHANGE]
Personal context: [PERSONAL NOTE]
Tone: [LETTER TONE]

Requirements:
- Start with a reference to the portfolio performance, not a generic greeting
- Acknowledge the benchmark comparison naturally without making excuses
- Briefly mention the portfolio activity in plain English
- Weave in the personal context naturally (1 sentence)
- End with a forward-looking statement about the next quarter priorities or a reminder to reach out with questions
- No specific investment predictions
- First names only — informal and warm
- Exactly 150 words

Part 3: Run your first batch (10 clients)

Open ChatGPT. For each client row in your spreadsheet:

1. Copy your master template prompt
2. Fill in the five variables from that client's row
3. Paste into ChatGPT
4. Copy the output into a Word doc or directly into your email platform

For 10 clients, this takes about 20 minutes — 2 minutes each.

Evaluate the quality. Are the letters distinct from each other? Do they sound personal? Does the personal context feel natural? If yes, you're ready to scale.

Part 4: Scale to your full client list

For your full list of 50-150 clients, organize by portfolio performance buckets first:

• Up significantly (above benchmark) — use "positive and forward-looking" tone
• Up modestly or in line with benchmark — use "encouraging" tone
• Down or below benchmark — use "steady/reassuring" tone

This lets you write 3 tone-appropriate master prompts instead of one generic one, and switch between them as you go through your list.

At pace, you can process 30-40 clients per hour. A 100-client batch takes 2.5-3 hours — versus a full day or more doing it manually. More importantly: advisors who skip personalized quarterly letters will now actually send them.

Part 5: Review and send

After batch generation, review 100% of the letters before sending — this is not optional. You're looking for:

• Any specific performance claims that aren't accurate
• Any personal notes that are awkwardly placed
• Any letter that sounds like every other letter (the AI should vary the phrasing)

Budget 1 hour of review for a 100-client batch. Then send from your email platform or CRM.

Real Example: End-to-End Workflow

Setup: Q4 is over. You export performance data from Orion — a spreadsheet with portfolio return for every account. You pull your CRM for personal notes from Q4 meetings.

Data table row for the Chen family:

• Client First Names: Michael and Joyce
• Portfolio Return: up 7.1%
• Benchmark Return: S&P was up 8.9% this quarter
• Notable Change: rebalanced from 65/35 to 60/40 per your guidance
• Personal Note: Joyce mentioned she's planning to retire in June
• Tone: encouraging (performed well, slightly under benchmark — explain the rebalancing)

Prompt filled in: "Write a personalized 150-word quarterly commentary for Michael and Joyce. Portfolio up 7.1%, S&P up 8.9%. We rebalanced from 65/35 to 60/40 this quarter. Joyce is planning to retire in June. Tone: encouraging — acknowledge the benchmark gap is partly explained by the rebalancing toward their retirement-ready allocation."

Output: "Q4 brought solid results for your portfolio — up 7.1% for the quarter, in a market where the S&P returned 8.9%. The slight difference is mostly by design: we moved your allocation from 65% stocks to 60% stocks this quarter, taking some risk off the table as Joyce's June retirement approaches. A portfolio positioned for income and stability will sometimes lag a pure equity benchmark in a strong quarter, but that's the right tradeoff for where you are in your planning. As we move into Q1, I'll be focused on finalizing your retirement income strategy and ensuring everything is in place before June. As always, reach out anytime if you'd like to talk through anything."

Time saved: 150 letters × 10 minutes each (manual) = 25 hours. With this workflow: 3 hours for generation + 1 hour review = 4 hours total. 21 hours saved every quarter.

What to Do When It Breaks

• Letters all sound the same → Add more variety to your personal notes column. The AI can only vary based on different inputs. If all your personal notes say "reviewing retirement," all letters will reference retirement similarly.

• Output is longer/shorter than 150 words → Add "Count the words before responding. Adjust to reach exactly 150 words" at the end of your prompt.

• AI includes specific performance predictions → Add a rule to your prompt: "Do not include any forward-looking performance predictions or market forecasts."

• Personal note sounds forced → Change to: "Naturally weave in the following personal context — it should feel like it belongs, not like it was inserted: [personal note]"

Variations

• Simpler version: Do this for just your top 20% of clients (by AUM) and send the rest a general market update. 80% of the value at 20% of the work.
• Extended version: Connect your Orion data export to a Google Sheet with an Airtable or Zapier automation that pre-fills your template — making the batch nearly instant.

What to Do Next

• This week: Run a 10-client test batch and evaluate quality
• This month: Build out your full client data table with personal notes
• Advanced: Explore whether your CRM can trigger automated draft generation when new portfolio data arrives

Advanced guide for financial advisor professionals. Always review AI-generated client communications before sending.