Overview

Effective prompts are the difference between a useful, accurate output and a long, noisy dump. This guide explains how to structure prompts for financial analysis so Claude returns precise, verifiable, and actionable results. It covers discovery (what data Claude can access), core prompting principles, handling complex analyses, common issues and fixes, and ready-to-use prompt patterns tailored for both Indian and global analysts.

1. Discover what data Claude can access

Before you start, confirm which integrations and datasets are available in your workspace. Different integrations provide different assets:

• Global examples: Daloopa (SEC filings/fundamentals), Kensho / S&P Global (market and relationship data).
• Practical Indian options (where direct integrations may be absent): Screener.in, Perplexity Finance, NSE/BSE filings, company investor presentations, local broker research (ICICI Direct, Motilal Oswal, Axis).
• Always begin sessions by asking Claude what it can access for the company or metric you need.

Suggested starter checks:

• Which S&P Global metrics are available through Kensho in this project?
• Can you retrieve segment-level revenue for Infosys from Screener.in or NSE filings?

These short checks save time and prevent re-work.

2. Core prompting principles

Be specific and precise

Bad prompt:

Claude, analyze Microsoft.

Good prompt:

Using Daloopa, retrieve MSFT revenue, operating margin, and free cash flow for Q1 2023–Q4 2024 and calculate YoY growth rates.

What to include: