Getting Started with Tableau Desktop

Tableau fundamentals, data connections, visualizations, dashboards.

Books worth reading

• Practical Tableau: Ryan Sleeper
• Tableau Desktop Certified Associate Exam Guide: Dmitry Anoshin et al. (also covers Tableau Server and Prep)
• Tableau Your Data!: Dan Murray (Tableau Desktop and Server)
• Ryan Sleeper’s other books on visualization

Tableau ships new features every quarter, but the fundamentals stay the same: dimensions, measures, a semantic model, chart types, live vs extract. Every BI tool has them, so once you know one tool well, the others come quickly.

The Surfalytics approach

The same Superstore dataset goes through the whole roadmap:

1. Module 1: a dashboard in spreadsheets.
2. Module 2: load it into Postgres and answer business questions with SQL.
3. Module 3: connect Tableau to the same data and build the dashboard there.

Superstore is Tableau’s own sample dataset. With one dataset you answer the same business questions in each tool and can compare the results.

Tableau products

• Tableau Desktop: where you connect to data and build views and dashboards. Windows and macOS (Power BI is Windows only).
• Tableau Server (and Tableau Cloud): where you publish, share and schedule.
• Tableau Prep: data preparation.
• Tableau Public: a free public server for portfolios. You can also download other people’s workbooks and see how they were built.
• Tableau Reader: a free app for opening workbooks someone shared with you.

Tips:

• If your company has Tableau Server, match your Desktop version to it. Without a server, install the latest one.
• A free one-year license is available through the student program. Members of Surfalytics have got it successfully. Coursera’s Tableau visualization course gives a 6-month license.

From data to dashboard

Tableau is the BI layer from the architecture diagram of Module 1: it sits on top of storage (files, Postgres, Snowflake, BigQuery and many more connectors).

Step	What it is
Connect	Pick a source: flat file, database or warehouse
Data source	Drag in tables and define joins. This is your semantic layer
Sheet (view)	One visualization built from dimensions and measures
Dashboard	Several views on one canvas, with shared filters
Story	A sequence of views, like slides, to tell a story

Dashboards in Looker, Superset or Metabase work the same way: build individual elements first, then combine them with shared filters.

Joins

Joins in Tableau are the same as in SQL, so BI developers must know SQL.

Join	Result
Inner	Only rows that match in both tables
Left	All rows from the left table, plus matches from the right (nulls where there is none)
Right	The opposite. In practice nobody uses it: swap the tables and use a left join
Full outer	Everything from both. Rarely useful for analytics and can produce confusing metrics, but handy for troubleshooting

Union is different: it stacks tables with the same columns, so you get more rows and the same columns. A join adds columns.

The data source tab also offers pivot and basic cleaning. Federated queries let you combine different sources (say Postgres and a CSV or Snowflake) in one data source: Tableau queries each separately and joins the results.

Live connection vs extract

	Live	Extract
How it works	Every change in the view generates a new SQL query to the database	One query when you create the extract, the result is cached in Tableau’s columnar engine (Hyper)
Data freshness	Always current	As fresh as the last refresh
Load on the database	Every user and every click	Only on refresh
Typical use	Small data, fast database	Many users, expensive warehouse (no Snowflake credits burned per click)

Under the hood, dragging and dropping in Tableau generates VizQL, which is converted to SQL (or MDX for cubes).

Don’t pull billions of raw rows into a BI tool. Live connections time out and extracts fail. BI needs aggregated data.

Dimensions and measures

The most important concept in BI:

• Dimensions are qualitative: year, month, product name, category. They add context. Blue in Tableau.
• Measures are quantitative: they can be counted and aggregated (sales, profit, quantity). Green in Tableau.

You can also group fields into folders, rename them, change data types (number, text, date) and create calculated fields.

Building a view

• Columns and Rows shelves, Filters, and Marks (color, size, label, tooltip).
• A date can be discrete (separate values) or continuous (a real time axis, which gives a line chart). Hold Alt/Option while dragging to choose the date part.
• Tableau picks a chart type for you, but you can change it. Never use a pie chart.
• Scope (table, pane, cell, line) and table calculations (running total, percent of total) are the tricky parts. They work like window functions in SQL. This is where official training pays off.
• Tableau suggests a blue/orange palette by default because red/green doesn’t work for color-blind readers (see the previous lesson).
• A filter applies to one view by default. Use Apply to Worksheets → All Using This Data Source to make it global.

Demo: first dashboard on Superstore

What the lesson builds:

1. Connect to the built-in Sample - Superstore Excel file (Orders, People, Returns) and join Orders to People.
2. Sales by Category: a sorted bar chart with labels and a relative-date filter (last two years) shown as a control.
3. Quantity by month: a continuous date as a line chart, split by category with color, labels only for min and max.
4. Profit by state: a map with a blue/orange legend.
5. Combine the three views in a dashboard: pick the size (there is a mobile layout), use layout containers, fit views to the whole area, add a title.

Even this basic dashboard lets a sales team see how categories perform and which states do well.

Tableau dashboards are not pixel-perfect. You drag and drop elements and tune the layout with containers.

Why Tableau (or any BI tool)

Pros	Cons
Automates routine work, fewer human errors	A longer learning curve, especially for people who have worked in spreadsheets for years
Easy to share insights, interactive even for non-technical users	Licensing cost
Creative: you can build almost any visualization

Two stories from the lesson:

• Daily dashboard in a training company. Updating a spreadsheet by hand took from 30 minutes to 3 hours every day. After rebuilding it in Tableau, it was one click and a PDF for the leadership team.
• Amazon WBR. In spreadsheets, every team spent 1–2 days preparing slides and one person combined them. In Tableau the WBR (with sparklines) became automated: with Tableau Server a schedule generates a PDF and emails it every Monday. It is faster, consistent and accurate, as long as data quality is solid (that’s Module 4).

If you apply for a BI role at Amazon, you will work with WBRs. Being able to explain how to build and automate one is a big plus.

Useful terms

Data blending is the older way to combine two sources. A primary source (for example actuals in a warehouse) and a secondary source (for example a forecast in a spreadsheet) are linked by month, date or product to get a fact-to-plan ratio.

File types

File	What it holds
Workbook (.twb)	Sheets, dashboards and stories
Data source (.tds)	Connection and semantic model
Data extract (.hyper)	The cached data itself. The heaviest one
Packaged workbook (.twbx)	A workbook with its data. Can be opened in Tableau Reader
Packaged data source (.tdsx)	A data source with its data, for sharing
Bookmark (.tbm)	A single saved sheet

How to practice

• Build a few projects on Superstore connected to your Postgres database, then publish them to Tableau Public.
• Look at the best dashboards on Tableau Public and download workbooks to see how they are made: fonts, background images, hyperlinks, containers.
• Makeover Monday: a weekly community challenge with a new dataset. Build your own version and compare it with others.
• The official Tableau Superstore dashboard is a much better example on the same data: KPI tiles, product and customer analysis, text-input what-if parameters, forecasting, trend lines and clustering.

Certification

There are Specialist, Associate and Professional exams for Desktop and Server. Preparing for one is a good way to learn up-to-date features, but the exam itself matters little: nobody checks, and you can list the certification name on your resume.

Your portfolio

A candidate who applies for a Tableau job without a Tableau Public profile is a red flag. The same goes for developers without GitHub. A hiring manager will also read your blog posts and check whether you put in real effort or had AI generate them.

Homework

1. Connect Tableau Desktop to your Postgres database with the Superstore data from Module 2.
2. Create a data source and choose live or extract. Be ready to explain why.
3. Build a few views and combine them into a dashboard. Use the official Superstore dashboard for inspiration.
4. Publish it to Tableau Public and add it to your resume.