Tableau Connecting Data Sources: Live vs Extract for Analysts

Analyst work starts before the first chart — at the connection. A wildlife-strikes CSV that opens with wrong date types or a 50M-row table queried live will break every visual that follows. This tutorial follows the exact workflow from script 01: establishing the connection, managing the Connections pane, verifying structure, and choosing performance options — so your foundation is correct before you design.

For the broader analyst journey, anchor this in the Data Analyst Roadmap. For modelling after connection, see Tableau Relationships vs Joins and Tableau Data Types and Roles.

What happens when you connect a file in Tableau?

Tableau opens the Data Source page as soon as you choose a connector — Text file for CSV or Microsoft Excel for workbooks Tableau Help: Connect to Data. You select the file from the start page (0:24 in the script demo), and Tableau loads the file list and preview grid. The orange Sheet 1 tab is your gate to the viz layer; nothing aggregates until you cross it.

Drag the wildlife-strikes table onto the canvas. Tableau infers headers from the first row and shows the row count immediately. If headers are missing, use the data interpreter or edit the header row in the preview.

How do you manage connections and tables?

The Connections pane on the left lists every file or database you have added Tableau Help: Connect to Data. Two controls matter for analysts:

• Edit Connection — fix a path change without rebuilding the workbook.
• Rename — give the source a stakeholder-friendly name before publishing.

Drag-and-drop adds tables to the canvas. For Excel, each sheet and named range appears separately — add only the tabs you need. Adding an unused 200k-row sheet inflates extracts and slows Hyper.

How do you verify data structure and types?

The preview grid's bottom pane shows column headers, assigned data types, and origin file — the metadata row the script highlights at 1:14 Tableau Help: Connect to Data. Check three things before Sheet 1:

1. Date fields parsed as Date, not String — otherwise timelines break.
2. Geographic fields assigned to Geography — otherwise maps fail.
3. Null vs empty — Tableau shows null as blank; treat it consistently later.

This step is also where data blending context appears: when you combine multiple sources, Tableau's type inference determines which field can be the linking field.

Should you use Live or Extract?

Live = real-time; Extract = snapshot Tableau Help: Extract Your Data. That single choice decides accuracy and speed:

// Conceptual: not a calculation, but the decision
// Live:  Dashboard -> Live SQL -> Source DB (fresh, slower)
// Extract: Dashboard -> Hyper (.hyper) -> Snapshot (fast, static)

Use Live when:

• Data changes intraday and decisions depend on it (support queues, inventory).
• Your database is a cloud warehouse (Snowflake, BigQuery) that handles concurrency.

Use Extract when:

• Source is a flat CSV/Excel with no refresh need.
• You need offline work or sub-second filtering on >1M rows via Hyper.
• You must reduce load on a production OLTP database.

Hyper extracts are columnar and memory-mapped, so a 10M-row CSV filters in milliseconds where Live would scan the source each interaction.

How do data source filters optimise performance?

Data source filters permanently exclude rows before they enter the workbook Tableau Help: Filter Data from Data Source. Find them at the top-right of the Data Source page (1:50 demo). Example: exclude Status = Test or Year < 2020.

Why this matters more than a sheet filter:

• Sheet filters still query excluded rows, then hide them.
• Data source filters shrink the extract and every query downstream.

Rule: if a row will never be analysed, filter it at the source. If a stakeholder might toggle it, use a context or dimension filter instead.

Quick reference and next steps

Next: learn how Tableau interprets those types in Tableau Data Types and Roles, then how it models multiple tables in Tableau Relationships vs Joins. Practice the workflow on a real CSV and publish one chart to your portfolio before moving on.

Educational note: Tableau vs Excel vs SQL choices here are for learning sequence, not a buying recommendation. The comparison capstone after 8 Tableau spokes will provide balanced guidance.