Recipes

Unsure which chart is best for your analysis? Here I demonstrate how to pull together many of the most frequently used charts used in business today, without using the Show Me function.

All the videos are running at their original 1x speed, and no sequences have been modified or shortened

Find the chart, have a play, then see how to build it


Dashboard not loading?

Download the workbook here: Most Requested Charts.twbx


These videos have no sound


1. Bar (0:05)

Bar charts are used to measure and compare categorical information.

Tableau can plot multiple measures against both a single hierarchical category or, multiple grouped categories or, as you will see further down, with the addition of information, you can also plot stacked and / or side-by-side charts too.

2. Side-by-side (0:07)

Cluster analysis:

Similar to bar charts, the side-by-side chart lets you compare multiple measures of dimensions or dimension members at the lowest selected grain, clustered in groups or cohorts suitable to your story or perspective.

3. Line (0:17)

Time series chart: track one or more measures over a time grain

4. Mixed Mode (0:15)

 

Often seen as a mixture of line and bar charts, the added benefit here can provide more insightful tracking analytics such as tracking actual to forecast and then, use bars to illustrate the % difference between the two points, which is especially useful if the gap is not immediately obvious.

Like line charts, mixed mode line-to-bar charts are plotted as continuous charts, but can only support a single dimension; it is not easy for example to create a side-by-side mixed-mode chart as Tableau will treat a multi-dimension chart as a discrete chart converting your lines into individual marks; I demonstrate how to achieve this in video 37.

5. Pie (0:10)

Discrete part-to-whole chart: Allows comparisons of a single measure and a single dimension.

Pie charts really should only be placed on summary or overview sheets and need to be limited to less than 10 (ideally less than 5) dimensions so where more dimensions are likely to be available, consider switching to use a Top 10.

Pie-charts can be awkward to see the comparison due to the width of the slice reducing as the slice reaches the centre; alternate suggestions would be:

  • stacked bar chars
  • doughnut charts
  • treemap's

6. Doughnut (0:39)

The same as a pie but with the centre portion removed.

Doughnut charts  do not have the centre portion thus, they are actually easier to understand than pie-charts as they do not use slices, the central section can be used for text or, if you are feeling particularly confident, a pie-chart perhaps for grouping the outer portion.

Unfortunately, doughnut charts are not native to Tableau and need to be built with a bit of trickery.

7. Stacked Bar (0:07)

 

Combination chart: Break-down the bar-chart values to the lower-grain components to understand contributions

8. Area (0:09)

Similar to the stacked bar chart but adds an additional dimension, perfect for time-based cohort analyses 

9. 100% Stacked Bar

As a stacked bar chart only each bar is broken down by percentage of total of their contribution, rather than their actual values.


10. 100% Stacked Area (0:28)

Exactly the same as the combination of the 100% Stacked Bar and, the area charts

11. Shape Tile (0:42)

Shapes can be useful for helping progress your story, or explaining the dimension in a word (and thus language) -free setting

 12. Lollipop (0:29)

Lollipop charts, nothing really more than a dot plot, but with the addition of a thinly-defined bar linking the dot to the y-axis to help your users identify the attribute to which the dot relates.

Can be thought of as a three-dimensional chart, looking at the flow of the dots demonstrates overall velocity, the bars, as a value and performance indicator, and the overall attribute as a indication of performance. 

13. Heatmap (0:09)

 

The classic method of bringing highlighting colour to a chart, helping you spot even the most subtle change in a sea of numbers.

Uses both size and colour so can be multi-dimensional.

14. Geo (Choropleth) (0:04)

Filled maps - choropleths. Another multi-dimensional chart, this time filling in the chart by geo hierarchy. Here I demonstrate colouring by state (United States), but at a different geo hierarchy, we could colour by region, city or town.

15. Scatter Plot (0:21)

The de-aggregate chart that plots the individual records on their own merit. Useful for measuring overall velocity, with the added benefit of a reference line, Tableau is able to provide your statistical reference information including the all-important p-value.

Extremely useful in identifying outliers and trend-setters.

16. Waterfall (0:31)

 

The exploded stacked bar-chart of the later 1990's, making a massive comeback since mid 2013. With colour, you are able to identify different perspectives.

As a cumulative chart, this helps the user understand each attributes contribution to the bottom-line (grand total), with negative values being given a level forum for their effect on the overall figure, as the next attribute shall begin its plot from the final position of the previous value.

Priceless in the efforts to demonstrate the absolute effect of a set of levers. 

17. Histogram (0:16)

Separates your values into bucket groupings, and then measures the count of records that fall into each bucket, eg, groups sales values by 10's of thousands of pounds: 0-10, 11-20, 21-30 etc

Useful for identifying, well anything you want really, just remember the y-axis attributes here will be formed of the discrete buckets from your measures rather than dimensions

18. Tile (General Purpose) (0:31)

Tiles can be useful as bordered windows, colour can be applied from a different dimension to the information reported allowing for multi-dimensional reporting: the information in the tile relative to a given perspective, the colour can reflect that perspective for a single dimensional analysis or, can be made of that across the whole dashboard, to provide that all-important cohort analysis.

19. Treemap (0:10)

Functionally the same as the heatmap only, the attributes are drawn together providing another level of meaning to how the data correlates to its neighbour, in a part-to-whole analysis guided and driven by both size and colour.

20. Word Cloud (0:27)

The word-cloud. Used less in business dashboards, and more for info-charts and marketing boards.

Dimension members as their actual identity, sized according to their measurable. often used as a more elaborate for of filter on interactive dashboards, or when trying to call-out a specific product or member, this chart should be used sparingly owing to the amount of space required to allow the chart to reach maximum impact.

21. Dumbbell (barbell) (0:41)

Dumbbell charts merely draw a connecting line between two related points to provide clarity of the relationship, this is especially useful when the difference between each point is hard to infer or worse, falls below expectation eg year-over-year for sales plotted monthly as in this demonstration: for almost all the data-points of ty plotting above py indicating improved sales performance, when one falls below the prior-year, a connecting line between the two points helps to indicate that these points are related.

22. Bubble (0:19)

Unlike word-clouds, this type of chart is rarely used, yet this can often provide a deeper clarity than the word-cloud.

Very similar to a word-cloud, perfect circles are drawn surrounding each member, the circumference of which is determined by the measure in use.

Use as an alternative to the pie or doughnut chart

23. Box & Whiskers (0:16)

The method described here utilises Tableau's immediate defaults, of 1.5 IQR (inter-quartile range) although these can easily be altered to the max/min of the sample, std-dev and percentile range.

Furthermore, for the purpose of time, this demonstrates the quickest method of building box (and whiskers) plots, although this is easy to achieve manually - a thorough understanding of table-calculations is required here.

Wiki provides a much clearer explanation than I: https://en.wikipedia.org/wiki/Box_plot


24. Shape (General) (0:35)

Mark each data-point with a shape, very useful where many data-points are to be plotted such that seeing subtle-differences could be easily missed

25. Pareto (1:41)

 

The 80-20 rule chart. Not always 80-20 but still, this chart helps understand where the greatest impact may be had, whether trying to identify who your most valuable customers are, or most effective employees, or where your greatest problems are.

Forget Top-n lists: Top-n is literally that, your Top-n items, but what happens if your Top-n is too wide or not wide enough?

This chart while provide you with that first-level identifier of just how wide you need to cast your net.

26. Period Over Period (1:19)

Use the methods employed in this video to create a period over period eg This Year vs Last Year chart.

27. Christmas Trees (2:46)

Demonstrates the classic up/down arrows for compact period over period display.

28. Bump (0:30)

 

Ranking: Classic comparison charts. Dimension rank of period

29. Bump with Callout (2:43)

Exactly the same as the bump chart, but with the added clarity of position for the selected dimension

30. Bar in Bar (0:49)

Similar to a stacked bar chart - indeed, this really is a stacked bar chart exploded so one measure is superimposed over the second.

Really only used for comparing two measures such as this year vs last year

31. Slope Chart (0:53)

 

Similar to a dumbbell chart, this provides a point-to-point analysis between two points - lit. the slope which can be quite powerful when combined with other marks: unlike a normal chart where we try to reduce the amount of marks, it is the mark breakdown that helps to provide context in the comparison

32. Control Chart (2:48)

Control charts help identify whether variations in your data fall outside your set standard deviation

33. Funnel (1:05)

The basic funnel typically used to illustrate the path taken by users and where they are likely to drop off.

The video demonstrates how to build a solid-looking funnel by switching to an area chart in the last couple of seconds, however, maintaining as a bar chart really helps you to understand where the greatest shrinkage happens

34. Gauge (6:43 // 3:20)

Gauges belong on machinery not dashboards, but, when your manager (or project sponsor) is adamant on their inclusion because "they look pretty cool" and / or "Excel can do them, and Microsoft is the voice of the people".

Here I walk you through their creation, and as ever, this is an unmodified Superstore Sales

Please ensure you exactly copy my actions here, this is a difficult chart to get right such that I have built this to make copying as easy as possible.

35. Expansive Drill (Brushing) (2:30)

Mostly used with bar charts, this isn't a new chart, but rather a means of drilling the selected item to a lower grain whilst keeping the rest of the chart at the higher grain.

The purpose is to allow users to see the breakdown of an attribute without losing focus on the remaining chart for example, a time bar chart at quarter-grain, enabling users to drill to the moths of Q2 (April, May, June), whilst keeping the rest of the chart at quarter:

Q1, April, May, June, Q3, Q4

You will need to create the action filter at dashboard level rather than worksheet level as demonstrated in the video as worksheet actions are not present at dashboard level 

36. Small Multiples (Trellis) (2:46)

Functionally the same as a line chart, this chart allows us to define a pane per attribute, which is useful in reducing chart clutter, especially when multiple measures need to be considered such as last period, current period and forecast.

Keeping the package together on the same axis, we can be sure that all axis are at the same scale to make comparison easy between each attribute.

Credit...

Special credit to the labelling method goes to Jeff Kritzman of Infectious Data 

https://infectiousdataviz.com/

which can be seen at:

https://infectiousdataviz.com/2019/01/11/tableau-tip-easy-elegant-small-multiples/

37. Lines with Side-by-Side Bars (6:05 // 2:20)

Analyse and compare measures from two or more cohorts such as sales performance vs forecast, for say East region vs West region; should be simple right? One of the most simplest charts that even Excel can build yet in Tableau, this is nigh impossible without a custom set.

38. Waffle (4:02)

Another part-to-whole chart, although this relatively smart chart will squash-up into really small 200 x 200 spaces making it ideal for when space is a premium, and, you can format the shape and colours and, the direction to read the chart: the one featured here reads bottom up.

Uses a custom scaffold as this chart tends to break when relying solely on the data and, being one of the highly customised charts, you shall need to make a chart per attribute

39. Shape Mapping (4:19)

Normally built using a hexagon, here I demonstrate how to build this simple chart using a circle shape that ships with vanilla Tableau instead - honestly, I actually think it looks better.

Switch the circles for weather and float over the top of another chart for weather effects.

This chart is much better when the co-ordinates are hardcoded though, as this shall break if the data is filtered. If needing to filter the data or use contexts, always build this chart from a duplicated set.

40. Radar (2:54)

For one or more attributes, instead of plotting side-by-side on a bar-plot, a radar is instead plotted around a circle which allows for easier understanding of relationships and distribution over time.
Think pie-chart and area-chart in one

41. Ratings (3:18)

For those moments when you want to visualise the outputs of those "on a scale of 1-10 how well did we do" questions. Tableau cannot natively build a star grid, but with the use of a 1-row/1-column update to the data, we can have a grid built in a jiffy, without harming the rest of the calcs throughout the workbook.

Just remember to HIDE rather than exclude the anchoring data

Holy Moly! (Bonus) (0:16)

Demonstrates the effect of adding trend-lines for every member of a given dimension. Sure, it looks like an accident on a paint factory but, looking harder, you can begin to get an idea of what a more prescriptive cohort analysis can harbour.