32 Advanced Tableau Charts and Custom Visualizations

32.1 Why Advanced Charts Matter

A bar chart answers most questions; the chart that answers a specific question, faster, is the one worth building.

Most business questions can be answered with a bar or line chart, and most dashboards should default to those. But certain questions reveal themselves more efficiently in a chart designed for them — a bullet chart for variance against target, a slope graph for two-period rank changes, a Pareto for cumulative contribution, a small-multiple sparkline grid for many time series at once.

The standard practitioner reference for this material is Practical Tableau by Ryan Sleeper (2018), which catalogues many of the patterns. For the broader vocabulary of advanced chart types, Jonathan Schwabish (2021) in Better Data Visualizations is the modern reference — covering more than 80 chart types organised by question.

For a visualisation-focused book, this chapter is where the chart-selection framework from Module 2 (Chapter 13) meets the BI tool that builds the charts. Tableau’s Show Me recommender, its dual-axis machinery, and its polygon-based custom-shape engine together cover virtually every chart the audience could need.

32.2 When to Move Beyond Bar and Line

A useful test before reaching for an advanced chart:

Is the question genuinely different from what a bar or line answers? If not, stay with the basic chart.
Will the audience read the advanced chart correctly without instruction? If not, either add a legend and short narrative or fall back to the basic.
Does the advanced chart compress information that would otherwise need multiple basic charts? If yes, the advanced chart usually wins.

The progression in many real dashboards: start with bar and line; replace specific bars with bullet charts when target comparison becomes routine; add small-multiple sparklines for monitoring views; introduce a single Pareto or waterfall for the executive summary. Advanced charts earn their place by question-fit, not by novelty.

32.3 Tableau’s Show Me — The Built-In Recommender

flowchart TD
    SM["Show Me<br>Recommender"]
    SM --> C["Comparison<br>Bar, side-by-side bar,<br>circle views, lines"]
    SM --> D["Distribution<br>Histogram, box plot,<br>scatter, density"]
    SM --> Co["Composition<br>Pie, treemap,<br>stacked bar, area"]
    SM --> T["Trend<br>Lines, dual lines,<br>area, sparklines"]
    SM --> R["Relationship<br>Scatter, scatter matrix,<br>bubble, packed bubble"]
    SM --> G["Geographical<br>Filled map, symbol map,<br>density map"]
    SM --> Sp["Specialised<br>Heatmap, highlight table,<br>tree map, Gantt"]
    style SM fill:#e3f2fd,stroke:#1976D2
    style C fill:#fce4ec,stroke:#AD1457
    style D fill:#fff3e0,stroke:#EF6C00
    style Co fill:#fff8e1,stroke:#F9A825
    style T fill:#e8f5e9,stroke:#388E3C
    style R fill:#ede7f6,stroke:#4527A0
    style G fill:#f3e5f5,stroke:#6A1B9A
    style Sp fill:#eceff1,stroke:#455A64

Tableau’s Show Me panel sits in the upper-right of the Worksheet view. Click any field in the Data pane and Show Me brightens the chart types that are compatible with the current selection.

The 24 chart types in Show Me cover the fundamental categories. For a richer chart, the analyst clicks past Show Me and builds it manually using the Marks card and Tableau’s mark types — Bar, Line, Area, Square, Circle, Shape, Text, Map, Pie, Gantt Bar, Polygon, Density.

32.4 A Catalogue of Advanced Charts

Comparison Charts

Chart	When to Use	Build
Bullet Graph	Single value vs target with banded reference	Show Me → Bullet, then add target as Reference Line on Detail
Bar in Bar	Two related measures (target vs actual) on same bar	Two measures dragged to Columns; second one slimmed and overlaid
Side-by-Side Bar	Group of categories repeated by sub-group	Show Me → Side-by-Side Bars
Lollipop Chart	Many categories with focus on the values	Bar chart + Circle on dual axis with synchronised scale
Dot Plot (Cleveland)	Many categories where bars feel heavy	Mark type → Circle on a horizontal axis
Slope Graph	Two time points across many categories	Line chart with only two points per line
Bump Chart	Rank evolution over time	Line chart of rank by period; mark rank with text

Distribution Charts

Chart	When to Use	Build
Box Plot	Distribution comparison across categories	Show Me → Box-and-Whisker
Violin (built)	Distribution shape and density together	Combine box plot with density mark on dual axis
Histogram	Distribution of one numeric variable	Show Me → Histogram
Density Plot	Smoothed distribution	Mark type → Density
Beeswarm	Many points with avoid-overlap jitter	Scatter with jittered position calculation
Pareto Chart	Cumulative contribution by sorted categories	Bar of values + line of running total of percent

Composition Charts

Chart	When to Use	Build
Stacked Bar	Composition with comparison of totals	Show Me → Stacked Bars
100 % Stacked Bar	Composition only, comparable totals	Right-click measure → Quick Table Calculation → Percent of Total
Treemap	Hierarchical composition with many small pieces	Show Me → Treemap
Donut Chart	Two-to-four category composition with central headline	Two pies on dual axis, smaller white-filled circle on top
Waterfall Chart	Decompose a starting to ending value	Gantt mark, with start and end calculated fields
Funnel Chart	Sequence with successive drop-off	Bar chart sorted by stage, with running difference labels

Trend Charts

Chart	When to Use	Build
Sparkline Grid	Many time series in one compact view	Small-multiple line charts with axes hidden
Dual-Axis Combo	Two related measures on one chart	Two measures on Rows; right-click second → Dual Axis; Synchronise Axis
Area Chart	Cumulative magnitude of one or a few series	Mark type → Area
Stacked Area	Composition over time	Same with multiple series stacked
Candlestick (Financial)	OHLC price data	Two Gantt marks per period: low-to-high stick, open-close box

Relationship and Specialised Charts

Chart	When to Use	Build
Scatter Plot	Relationship between two numeric variables	Show Me → Scatter
Scatter Matrix	Pairwise relationships across many variables	Build with Measure Names on Columns and Rows
Bubble Chart	Three-variable comparison (X, Y, Size)	Scatter with Size on Marks
Packed Bubbles	Categorical comparison with size encoding	Show Me → Packed Bubbles
Word Cloud	Frequency of categorical text values	Show Me → Word Cloud
Heatmap	Two-variable colour-encoded matrix	Mark type → Square, Color = measure
Highlight Table	Numeric table with conditional colour	Show Me → Highlight Table
Sankey Diagram	Flows between categorical states	Polygon mark with densification (advanced)
Radar / Spider	Multi-attribute profile	Polygon mark with calculated angle and radius
Density Map	Geographic clusters of points	Mark type → Density on map
Symbol Map	Point-level locations	Mark type → Circle on map
Filled Map (Choropleth)	Region values	Show Me → Filled Map
Gantt Chart	Project tasks with durations	Show Me → Gantt

32.5 Building Dual-Axis and Combination Charts

The Dual-Axis Chart is the single most-used advanced technique in Tableau. It allows two measures to share an axis, often with one drawn as a bar and the other as a line.

To build:

Drag two measures to Rows (or Columns).
Each measure produces its own pane.
Right-click the second measure pill → Dual Axis.
The two panes collapse into one. Each measure now has its own y-axis on opposite sides.
Right-click either axis → Synchronize Axis to force the two scales to share the same range. Crucial — without it, the chart silently misleads.
On the Marks card, Tableau gives each measure its own Marks panel; change one to Bar and the other to Line for a combination chart.

The dual-axis pattern is the foundation of bullet graphs (bar + reference line), bar-line combos (actual + target), donut charts (pie + smaller pie), and many polygon-based custom visualisations.

32.6 Custom Visualisations with Polygon and Custom Shapes

When the chart the analyst needs does not exist in Show Me, Tableau’s polygon mark and custom-shape capability fill the gap.

Polygon Mark: Each row of the data is treated as a vertex; Tableau connects vertices in order to form a closed shape. With suitable data preparation (densification — adding extra rows to define the shape), the polygon mark builds Sankey diagrams, radar charts, sunburst charts, and anything else expressible as a closed-form polygon.

Custom Shapes: Image files (PNG, GIF) can be loaded into Tableau’s Shapes repository and used as marks. Useful for icon-based KPIs, country flags (subject to the cultural cautions from Chapter 19), product images, or branded marks.

Background Images: A floor plan, schematic, or process diagram can be loaded as a background image, with marks placed on calculated coordinates to overlay data on the underlying image.

These techniques are advanced and usually require a custom data-preparation step. Ryan Sleeper’s Practical Tableau contains step-by-step recipes for the most-asked patterns.

32.7 Tableau Extensions and Viz Extensions

Beyond Tableau’s built-in capabilities, two extension models add chart types and features:

Dashboard Extensions: Web-based panels embedded in a dashboard. Examples: write-back to a database, advanced filtering UIs, data dictionaries.
Viz Extensions (announced 2024): Custom chart types built using a JavaScript SDK and used like any other mark. Examples: Sankey diagrams, sunburst charts, network graphs.

The Tableau Exchange marketplace lists hundreds of free and paid extensions. For analysts in larger firms, vetted internal extensions can be deployed on Tableau Server / Cloud with controlled access.

32.8 Best Practices for Advanced Charts

Default to bar and line; reach for advanced charts only when the question genuinely demands it.
Annotate generously: A bullet chart, a Pareto, or a slope graph reads quickly only with clear labels and a short title that names what the chart shows.
Honour Cleveland’s hierarchy: Even in advanced charts, encode the most important quantity in position or length, not in colour or area.
Synchronise dual axes: Almost always; the unsynchronised dual-axis chart is one of the most misleading patterns in BI.
Limit the chart vocabulary: A dashboard with twelve different advanced chart types is harder to read than one with three. Pick a small palette of advanced charts and use them consistently.
Test on a non-analyst: Show the advanced chart to someone who does not work with charts daily; if they cannot read it within ten seconds, simplify or annotate.
Document the build: Advanced charts are harder to maintain. A short note in the worksheet description (right-click sheet tab → Edit Description) helps the next analyst.

32.9 Common Pitfalls

Unsynchronised Dual Axis: Two measures on different scales producing whatever correlation the designer chose by axis range.
Pie Chart with Eight Slices: Pies barely work for three to four categories; donut and treemap are the better alternatives for more.
Word Cloud as Quantitative Encoding: Word clouds encode by area and angle of text; a sorted bar chart is almost always clearer.
Radar Chart with Twelve Axes: The polygon’s area becomes meaningless past five or six axes; comparison breaks down.
Sankey Without Validation: A Sankey built from polygon marks with the wrong densification produces visually plausible but numerically wrong flows.
Custom Shape Overuse: Replacing every category bar with a custom icon; the icons compete for attention and overall comparison becomes harder.
Treemap of Flat Data: Treemaps need a hierarchy; using one for flat categorical data is less clear than a sorted bar.
Show Me as Default: Reaching for whatever Show Me suggests instead of starting from the audience’s question.
Funnel for Non-Funnel Data: Funnel charts visually imply attrition; using one where the data has no attrition is misleading.

32.10 Illustrative Cases

A Bullet Graph Replaces a Cluttered Dashboard

A monthly performance dashboard with six bar-and-target side-by-side charts is replaced by six bullet graphs. The footprint shrinks by half; readers find the variance instantly. The bullet graph is the canonical example of an advanced chart that compresses information without sacrificing clarity.

A Pareto Reveals the 80-20 in Inventory

An inventory analyst plots SKU contribution to revenue using a Pareto chart. The cumulative line crosses 80 % of revenue at SKU number 240 of 2,800 — the classic 80/20 pattern, instantly visible. The same finding from a sorted bar chart would have required the reader to do the cumulative arithmetic.

A Sankey Maps a Customer Journey

A digital firm’s customer-journey analysis is rendered as a Sankey diagram from Channel to Landing Page to Action. Band widths show the volume at each step; the most leaked transition becomes visible at a glance.

32.11 Hands-On Exercise: Building Advanced Tableau Charts

Aim: Build five advanced charts in Tableau — a bullet graph, a Pareto, a dual-axis combo, a small-multiple sparkline grid, and a slope graph — using a single retail dataset.

Deliverable: A Tableau workbook (.twbx) with five worksheets, one per chart type, plus a short comparison page showing the same data rendered as a basic bar for contrast.

32.11.1 Step 1 — Sample Data

The same sales.csv from Chapter 31 (or any retail transaction file). Connect via Tableau Desktop’s Get Data → Text File.

32.11.2 Step 2 — Bullet Graph

Drag Region to Rows and SUM(Sales) to Columns.
Click Show Me → Bullet Graph.
Tableau auto-builds with the second measure becoming the reference. If targets are stored in a separate field (Target Sales), drag both measures and let Show Me pair them.
Right-click the axis → Edit Reference Line to format the bands (acceptable, good, excellent ranges).

The bullet graph compresses value, target, and performance bands into a chart half the height of three separate bars.

32.11.3 Step 3 — Pareto Chart

Drag Product to Columns and SUM(Sales) to Rows.
Sort Product in descending order of SUM(Sales).
Add a second SUM(Sales) to Rows.
Right-click the second measure → Quick Table Calculation → Running Total.
Right-click again → Add Secondary Calculation → Percent of Total.
Right-click the second pill → Dual Axis and Synchronize Axis.
Change the second mark to Line; the first remains Bar.

The chart shows the bar of values plus the cumulative-percentage line, with the 80 % crossing point visible.

32.11.4 Step 4 — Dual-Axis Combo (Actual vs Target Bar-Line)

Drag Month to Columns, SUM(Sales) to Rows, then SUM(Target) also to Rows.
Right-click the second measure pill → Dual Axis.
Right-click the secondary axis → Synchronize Axis.
On the Marks card, set the first measure’s mark type to Bar and the second’s to Line.
Format the line to be thin and a contrasting colour.

The chart is the canonical bar-line combo — actual bars with target line.

32.11.5 Step 5 — Small-Multiple Sparkline Grid

Drag Region to Rows and Month to Columns.
Drag SUM(Sales) to Rows as a measure.
Right-click the resulting line chart’s axis → Edit Axis → uncheck Include zero and reduce header size.
Right-click the axis again → Show Header to off; do the same for the y-axis.
Format each region line cleanly; the result is a grid of small line charts, one per region.

The sparkline grid shows trend across regions in a compact form.

32.11.6 Step 6 — Slope Graph

Filter the data to two years — say 2024 and 2026.
Drag YEAR(Order Date) to Columns, SUM(Sales) to Rows, and Region to Color on the Marks card.
The result is a multi-line chart with two points per line.
Format: thin lines, no axis numbers, large category labels at the right.

The slope graph compresses change between two periods, across many categories into one easy-to-read view.

32.11.7 Step 7 — Compose the Demonstration Workbook

Build a one-page comparison showing the same data rendered as:

A basic bar chart of region sales.
A bullet graph of region sales vs target.
A Pareto of products by sales.
A dual-axis combo of sales vs target by month.
A sparkline grid of regional sales over time.
A slope graph of region sales 2024 vs 2026.

Place all six on a dashboard with a short title for each and a one-line caption explaining what question each answers. The contrast between the basic bar and the five advanced charts demonstrates when each advanced chart earns its place.

32.11.8 Step 8 — Connect to the Visualisation Layer

Advanced charts in Tableau are not exotic for their own sake. Each one is a tool fitted to a specific question:

The bullet graph answers am I on target in a fraction of the space.
The Pareto answers which few items drive most of the value.
The dual-axis combo answers how does actual track target over time.
The sparkline grid answers how do many things trend at once.
The slope graph answers which categories rose or fell between two periods.

The dashboard that mixes basic and advanced charts judiciously delivers more insight per square centimetre than one dominated by either alone.

Files and Screen Recordings

Tableau workbook (yuvijen-advanced-charts.twbx), the source sales.csv, and screen recordings of each chart type being built will be embedded here.

Summary

Concept	Description
Foundations
Why Advanced Charts Matter	A bar chart answers most questions; the chart that answers a specific question faster is the one worth building
When to Move Beyond Bar and Line	Use advanced charts only when the question genuinely differs from what bar or line answers
Show Me and Mark Types
Show Me Recommender	Built-in panel that brightens compatible chart types based on selected fields
Mark Types	Bar, Line, Area, Square, Circle, Shape, Text, Map, Pie, Gantt Bar, Polygon, Density
Comparison Charts
Bullet Graph	Single value vs target with banded reference; built via Show Me Bullet
Bar in Bar	Two related measures on the same bar; built via two measures with one slimmed and overlaid
Side-by-Side Bar	Group of categories repeated by sub-group; built via Show Me
Lollipop Chart	Many categories with focus on values; bar plus circle on dual axis
Cleveland Dot Plot	Many categories where bars feel heavy; circle mark on horizontal axis
Slope Graph	Two time points across many categories; line chart with two points per line
Bump Chart	Rank evolution over time; line chart of rank by period
Distribution Charts
Box Plot	Distribution comparison across categories; built via Show Me Box-and-Whisker
Violin Plot	Distribution shape and density together; box plot combined with density mark
Histogram	Distribution of one numeric variable; built via Show Me Histogram
Density Plot	Smoothed distribution; mark type Density
Beeswarm	Many points with avoid-overlap jitter; scatter with jittered position calculation
Pareto Chart	Cumulative contribution by sorted categories; bar plus running-total line
Composition Charts
Stacked Bar	Composition with comparison of totals; built via Show Me Stacked Bars
100 Percent Stacked Bar	Composition only with equal totals; quick table calculation Percent of Total
Treemap	Hierarchical composition with many small pieces; built via Show Me
Donut Chart	Two-to-four category composition with central headline; two pies on dual axis
Waterfall Chart	Decompose a starting to ending value; Gantt mark with start and end calculated fields
Funnel Chart	Sequence with successive drop-off; bar chart sorted by stage with running difference labels
Trend Charts
Sparkline Grid	Many time series in one compact view; small-multiple lines with axes hidden
Dual-Axis Combo	Two related measures on one chart; second measure on dual axis with synchronised scale
Area Chart	Cumulative magnitude of one or a few series; mark type Area
Stacked Area	Composition over time; multiple series stacked Area
Candlestick	OHLC price data; two Gantt marks per period
Relationship and Specialised Charts
Scatter Plot	Relationship between two numeric variables; built via Show Me Scatter
Scatter Matrix	Pairwise relationships across many variables; Measure Names on Columns and Rows
Bubble Chart	Three-variable comparison X Y and Size; scatter with Size on Marks
Packed Bubbles	Categorical comparison with size encoding; built via Show Me
Word Cloud	Frequency of categorical text values; built via Show Me
Heatmap	Two-variable colour-encoded matrix; mark type Square with Color measure
Highlight Table	Numeric table with conditional colour; built via Show Me Highlight Table
Sankey Diagram	Flows between categorical states; polygon mark with densification
Radar Chart	Multi-attribute profile; polygon mark with calculated angle and radius
Density Map	Geographic clusters of points; mark type Density on map
Symbol Map	Point-level locations; mark type Circle on map
Filled Map	Region values; built via Show Me Filled Map
Gantt Chart	Project tasks with durations; built via Show Me Gantt
Dual-Axis and Custom Visualisations
Synchronize Axis	Force two dual-axis scales to share the same range; non-negotiable for honest dual-axis charts
Polygon Mark	Each row treated as a vertex; Tableau connects vertices to form closed shapes for custom visuals
Custom Shapes	Image files used as marks; useful for icon-based KPIs and branded marks
Background Images and Extensions
Background Images	Floor plans, schematics, or process diagrams overlaid with marks on calculated coordinates
Dashboard Extensions	Web-based panels embedded in a dashboard; write-back, advanced filters, data dictionaries
Viz Extensions	Custom chart types built via JavaScript SDK and used like any other mark
Best Practices
Default to Bar and Line	Default to bar and line; reach for advanced charts only when the question demands it
Annotate Generously	Annotate advanced charts generously with labels, titles, and short captions
Honour Cleveland Hierarchy	Even in advanced charts, encode the most important quantity in position or length
Synchronise Dual Axes	Almost always synchronise dual axes; unsynchronised is one of the most misleading patterns
Limit Chart Vocabulary	A dashboard with twelve different advanced chart types is harder to read than one with three
Test on Non-Analyst	Show the advanced chart to a non-analyst; if unreadable in ten seconds, simplify or annotate
Document the Build	A short note in the worksheet description helps the next analyst maintain advanced charts
Common Pitfalls
Unsynchronised Dual Axis	Pitfall of two measures on different scales producing designer-controlled implied correlation
Pie with Many Slices	Pitfall of pie charts with eight slices; donut or treemap is clearer
Word Cloud as Quantitative	Pitfall of word clouds encoding by text area and angle when sorted bar would be clearer
Radar with Many Axes	Pitfall of radar charts past five or six axes where polygon area becomes meaningless
Sankey Without Validation	Pitfall of polygon-mark Sankey with wrong densification producing numerically wrong flows
Custom Shape Overuse	Pitfall of replacing every category with a custom icon and breaking overall comparison
Treemap of Flat Data	Pitfall of treemap on flat categorical data without hierarchy
Show Me as Default	Pitfall of reaching for whatever Show Me suggests instead of starting from the question
Funnel for Non-Funnel Data	Pitfall of using a funnel where the data has no attrition