Data Overview

Azure Blob Storage doesn't have predefined entities like a CRM or API. Instead, you sync files from your container. The data structure depends entirely on the format and content of your blobs.

File formats and structure

Structured formats

Format	Row/record structure	Data types	Use case
CSV	Comma-separated columns, first row = headers	Text, numbers (as strings unless converted)	Sales transactions, customer lists, inventory
JSONL	One JSON object per line	Nested objects, arrays, mixed types preserved	API logs, event streams, user activity
Parquet	Binary columnar format	Native types (int, float, string, date, etc.)	Analytics data, large datasets, BI tools
Excel	Rows and columns in sheets	Mixed types per column	Budgets, forecasts, reports

Unstructured format

Format	Content	Example use
Unstructured	Raw text, logs, markdown, HTML	Log files, error messages, documentation

Common import patterns

Single file sync — Pull one CSV or Excel file into Google Sheets for analysis.

Multi-file append — Use glob patterns like monthly/*.csv to match all monthly reports and Append them into a single BigQuery table. This creates a historical archive without duplicate logic.

Nested folder sync — Glob patterns like data/**/*.parquet sync Parquets from all subfolders in the data directory, useful for tiered storage structures.

Format conversion — Export data from a legacy system as Excel, sync to Coupler.io, and land it in BigQuery as a proper data type-aware table.

Use cases by role

Data analysts

Sync Parquet or CSV files from data lakes into BigQuery or Snowflake. Use glob patterns to match versioned exports (e.g., exports/v2/*.parquet) and Append transformations to build historical tables for trend analysis and modeling.

Finance teams

Data engineers

Operations

Platform-specific notes

• Glob patterns are powerful — *.csv matches files in the root; **/*.csv matches nested folders; data/2024/*.parquet targets a specific year
• Start date filters by modification time — Files touched after the start date are synced; unchanged files are skipped on refresh
• Excel syncs one sheet — If your workbook has multiple sheets, create separate data flows for each or export individual sheets as CSVs
• JSONL requires line-delimited JSON — Each line must be a complete, valid JSON object (not pretty-printed)
• Large files and performance — Parquet files compress better than CSV; for very large datasets, consider splitting into multiple containers or using date-based folder structures