Student Guide
Module 07
Documents, Data, and Artifacts
Working with files: CSVs, contracts, handbooks, financial data. A messy 50-row CSV that takes an hour to analyze by hand. Claude finds the anomalies in under a minute, if you ask the right way.
Charter Oak Strategic Partners · Claude Mastery Program
- How to upload and analyze files: CSVs, documents, financial data
- The difference between a vague prompt and a structured analysis prompt
- How Claude finds data quality issues without being told what to look for
- How to use Claude for document analysis, variance analysis, and data auditing
File-Based Workflows
Claude reads uploaded files: CSV data, text documents, PDFs, code files, images. This module introduces the work that occupies most business hours. Analyzing a messy 50-row CSV by hand takes an hour. Claude does it in under a minute. Reading a 20-page contract for dollar amounts takes 45 minutes. Claude does it in 30 seconds.
“What can you tell me about this data?” produces a description of columns.
“Identify the three largest revenue drivers, flag data quality issues, and format the output as an executive briefing” produces an analysis you can send to your boss.
Same file. Same tool. Different prompt. Different result.
Three Types of File Analysis
Upload a CSV with intentional quality issues: blank fields, mixed deal statuses, unknown customers, discounted prices, missing rep names. Claude finds all five issues when asked to audit.
Key: Tell Claude to look for problems. If you assume the data is clean, Claude will too.
Upload a handbook excerpt. Claude identifies legally risky policies: the no-call/no-show resignation clause, the jury duty compensation requirement, the broad electronic monitoring authority, the undefined “reasonable suspicion” drug testing criteria.
Key: A junior HR associate might take two hours. Claude takes 30 seconds.
Upload a P&L CSV. Ask Claude to write a variance analysis for the CFO. Claude calculates revenue variance (~$600K miss), identifies margin compression (42% budget vs. 37.6% actual), connects headcount shortfall to capacity, and writes the narrative a CFO expects.
Key: Specify the audience (CFO), the format (variance narrative), and the scope (Q4 vs. budget).
Exercise: The Data Detective
Instructions: Upload the messy CSV and write your own analysis prompt. The challenge:
- Find all the data quality issues
- Calculate the key metrics (total revenue, top rep, segment performance)
- Identify the three most important trends
Scoring: Did you catch the missing values? The unassigned rep? The negotiated discount? The Pipeline deal that should not be in the revenue total?
Reflection
Questions to Consider
- What files do you work with every week that Claude could analyze faster?
- Where does your team spend the most time on data cleanup or document review?
- How would you prompt Claude differently for a data audit versus an executive summary?
Supported Files
CSV, text, PDF, code, images. Upload-based (Cowork in Module 10 adds direct file access).
Prompt Quality = Analysis Quality
“Analyze this” = column descriptions. Specific instructions = work you can send upstairs.
Data Quality
Claude handles messy data without pre-cleaning. But tell it to look for problems. Assumption of cleanliness is inherited.