02 · Connecting to Data Sources

Every Tableau workbook starts on the same screen: the Connect pane. This module covers the connection types you'll use most, the Data Source page where you shape data before analysis, and the live-vs-extract decision introduced in Module 1 — using the Orders table from Module 1 as the worked example throughout.

1. The Connect pane

1. Opening Tableau Desktop lands you on the Start screen, split into a Connect panel (left) and a list of recent/sample workbooks (right).
2. Under To a File, you'll see options including Microsoft Excel, Text file (CSV), JSON file, and PDF file — the most common entry point for a beginner working with an exported spreadsheet.
3. Under To a Server, you'll see database and warehouse connectors — Microsoft SQL Server, Snowflake, Google BigQuery, PostgreSQL, and dozens more — each requiring server address and credentials rather than a local file path.
4. For this course's Orders dataset (Module 1), you'd pick Microsoft Excel or Text file if it were saved as a spreadsheet/CSV — the connection type Tableau treats as the simplest, with no server round trip needed.

2. The Data Source page

After connecting, Tableau opens the Data Source page — a canvas for shaping the data before it reaches a worksheet.

1. The top-left lists available tables/sheets from your connection; drag one onto the canvas to add it.
2. Dragging a second table near the first prompts Tableau to draw a relationship (the modern default, noeditable double line) or you can explicitly build a join (inner/left/right/full, a single-line connector) — the difference and when to use each is covered in Level 2 Module 3, since Level 1 uses a single Orders table.
3. The bottom half previews the actual rows — use this to sanity-check that column names and types look right before building any view. For Orders, you'd confirm here that Order Date is recognized as a Date type (calendar icon) and Sales/Quantity/Profit as Number types (# icon), not Text — a common import problem when a source file has stray formatting like currency symbols in a numeric column.
4. Right-click any column header in the preview to Rename, Change Data Type, or Hide — fixing a wrongly-typed column here (e.g. a Sales column that imported as Text because of a $ prefix) saves having to fix it later inside every worksheet.

3. Live vs. Extract, in practice

The Live/Extract toggle sits at the top-right of the Data Source page, and again at the top of the workbook once you're on a worksheet.

1. Live stays selected by default for most file and database connections. Every filter, sort, or new field triggers a live query against the source.
2. To create an Extract, click Extract (top right) then Edit... to optionally add filters (e.g. only pull the last two years of orders) before Tableau builds the .hyper file, or just click the toggle and let it pull everything.
3. Once extracted, use Data > Extract > Refresh to pull updated source data into the existing extract on demand — on Tableau Server/Cloud, this refresh can instead be scheduled to run automatically (Level 2 Module 9).
4. Practical rule of thumb for this course's small Orders dataset: either works identically since the whole table is 8 rows, but in real usage, prefer Extract for a file-based source you don't expect to change during your session — it makes dashboards feel snappier since Tableau isn't re-reading the file on every interaction.

4. Joining vs. a single flat table

Level 1 uses Orders as one flat table specifically so you can focus on visualization mechanics first. In real datasets, Sales data is often split across multiple tables (an Orders table and a separate Customers or Products table linked by an ID) — Tableau calls combining them a relationship or a join, both configured on the Data Source page exactly as described in Section 2. This is covered in depth in Level 2 Module 3, once you've built enough single-table visualizations to appreciate why splitting data across tables matters.

How It Actually Works

"Connecting" to data doesn't hand Tableau a static copy — it registers a metadata description (column names, inferred types, and, for Extract, a materialized copy) that VizQL's query generator reads before it can build anything:

1. On a Live connection, the Data Source page's schema (field names, types) is used purely to build correct SQL — every worksheet action still round-trips to the real source. Changing Sales from Text to Number here (as in the Exercise) doesn't change a single byte in the source database; it changes the CAST/type-coercion Tableau applies to that column when it writes the generated query, so SUM([Sales]) becomes a valid numeric aggregate instead of a string-concatenation error.
2. On an Extract, connecting triggers Tableau to run one pass over the source, apply any type coercions and initial filters from the Data Source page, and write the result into a .hyper file — a columnar, compressed on-disk format. From that point forward, every worksheet query is compiled and executed against Hyper locally, not the original source, which is why an extract's numbers can drift from a live database until you explicitly refresh (Data menu → Extract → Refresh) — you're re-running that same import pass, not re-pointing a live pipe.
3. Joining a second table (mentioned at the end of this module, expanded in Level 2 Module 3) changes the FROM/JOIN clause VizQL generates for every subsequent query, even ones that only reference fields from the original table — this is the mechanical reason a badly-chosen join type (e.g. an inner join that silently drops unmatched Order IDs) can make totals computed later in the course come out lower than a hand calculation against Orders alone expects.

Cheat sheet

Concept	Where / How
Connect to a file	Start screen → "To a File" → Excel/Text/JSON/PDF
Connect to a database	Start screen → "To a Server" → pick connector
Shape data pre-analysis	Data Source page (tables canvas + row preview)
Fix a column's type	Right-click header in preview → Change Data Type
Toggle Live/Extract	Top-right of Data Source page or worksheet
Refresh an extract	Data menu → Extract → Refresh
Combine multiple tables	Drag a second table onto Data Source canvas

Exercise

Imagine the Orders table (Module 1) was imported from a CSV where the Sales column values were formatted as $1,200.00 and imported as a Text type instead of Number. Write down, step by step, exactly which pane you'd open and which menu commands you'd use to (1) confirm the column is mistyped, and (2) fix it so Sales becomes a proper numeric field ready for aggregation (SUM) in a worksheet.