dataviz/docs/04-express.md

# EXPRESS – Choose proper visualization

EXPRESS covers all aspects of choosing the proper visualization in reports and
presentations.

_Proper visualization_ means that reports and presentations contain charts and
tables, which convey the desired message along with the underlying facts as
quickly as possible.

This chapter covers utilizing the correct types of charts and tables, replacing
inappropriate visualizations and representations, adding comparisons, and
explaining causes.

1. [Use appropriate object types](#ex-1-use-appropriate-object-types)
2. [Replace inappropriate chart types](#ex-2-replace-inappropriate-chart-types)
3. [Replace inappropriate representations](#ex-3-replace-inappropriate-representations)
4. [Add comparisons](#ex-4-add-comparisons)
5. [Explain causes](#ex-5-explain-causes)

## EX 1 Use appropriate object types

Choosing the appropriate _object type_ is of prime importance for the
comprehension of reports and presentations.

We use tables when looking up data. Tables have a high information density. They
are clear, they are honest, they do not want to highlight, and they typically do
not want to visually convey a certain message. So they do not compete with
charts.

Charts on the opposite are always biased. It is the selection of data, the
selection of the chart type, and the usage of highlighting which makes the
difference. We evaluate charts by asking whether they transfer the intended
message effectively and in a proper way. So charts cannot be replaced by tables.

The following section is about choosing the appropriate types of charts and
tables. It presents in detail different types, layouts, and examples as well as
their proper application.

## EX 1.1 Use appropriate chart types

![Figure EX 1.1: Use appropriate chart types](img/ex-1.1.png)

A _chart_ is a graphical object, in which visualization elements such as
columns, bars, and lines represent data.

This section discusses the types, layout, and examples of _single charts_.
_Overlay charts_ _and multiple charts_ are discussed in the CO 4 “[Add
elements](06-condense.md#co-4-add-elements)” and CO 5 “[Add
objects](06-condense.md#co-5-add-objects)”.

The most important groups of business charts are those showing development over
time (charts with horizontal category axes), those showing structural
relationships (charts with vertical category axes), and those showing x‑y
charts, scatter plots, and bubble charts (charts with two value axes), see
Figure EX 1.1.

Other chart types are of lesser interest in business communication and will be
treated in a later version of the standards.

![Figure EX 1.1-1: Chart Types](img/ex-1.1-1.png)

Looking at charts with horizontal and vertical category axes, the chart
selection matrix displayed in the figure aids in selecting the right chart type
for time series and structure analyses.

In the following sections, the correct usage of _charts with horizontal category
axes_, _charts with vertical category axes_, and *charts with two value axes* is
discussed in greater detail.

**Charts with horizontal category axes**

Charts with horizontal category axes (short: _horizontal charts_) typically
display time series. Use the horizontal category axis as a time axis.
Vertically, the visualization elements represent the data per time period or
point of time (there is no need to show a vertical value axis as the
visualization elements carry their own values). Time category axes run from left
to right and show characteristics of period types (e.g. months or years) or
points of time (dates).

In general, the data series of a _horizontal chart_ is represented by columns
(single, stacked, grouped), vertical pins, horizontal waterfalls, or lines.
_Vertical pins_ can be considered very thin columns. Because of their
importance, they are dealt with in a separate section.

Here follows the grouping of _horizontal chart types_:

**Single column charts**

![Figure EX 1.1-2: Single column charts](img/ex-1.1-2.png)

In general, _single column charts_ (short: single columns) are used to display
the temporal evolvement of one data series.

Single columns consist of:

-   **Horizontal category axis:** The _horizontal category axis_ represents with
    its labels the respective time periods or points of time. The part on
    “Semantic rules” suggests to use the category width (see width A in the
    figure) for identifying the period type (see UN 3.3 “[Unify time
    periods](09-unify.md#un-33-unify-time-periods-and-points-of-time)”).

-   **Columns**: One _column_ per time period or point of time extends from the
    category axis in accordance with the respective value. Columns are displayed
    in the foreground of the category axis, so that the length of the column is
    not hidden. The part on “Semantic rules” suggests that the ratio of column
    width to category width (see ratio B/A in the figure) represents information
    about the measure type (see UN 3.1 “[Unify
    measures](09-unify.md#un-31-unify-measures)”).

-   **Legends**: As there is only one data series, the legend (name of the data
    series) is part of the chart title.

-   **Data labels**: _Data labels_ name the values of the data series
    corresponding to the length of the respective columns. Position the labels
    of positive values above their respective columns, the labels of negative
    values below.

**Stacked column charts**

![Figure EX 1.1-3: Stacked column charts](img/ex-1.1-3.png)

_Stacked column charts_ (short: stacked columns) represent more than one data
series (e.g. multiple products, countries, or divisions), see the figure on the
left.

Stacked columns consist of:

-   **Horizontal category axis:** See single column charts.

-   **Columns**: The columns (see single column charts) are divided into
    segments (Excel names them “data points”) representing the data series
    (stacked columns).

-   **Legends**: Legends (names of the data series) are positioned either on the
    far left side with right-aligned text or on the far right side with
    left-aligned text. The column segments define their vertical position,
    centered vertically with the data labels of the respective column segment.
    If a segment at the far left or far right is missing or has a very small
    size, its legends might need assisting lines.

-   **Data labels**: _Data labels_ name the values of the data series
    corresponding to the length of the respective column segments. If the sum of
    the column segments of a category is positive (column pointing upward), the
    label of the sum is positioned above the respective column, if negative
    (column pointing downward), it is positioned below.

It must be pointed out that stacked columns should only be used if all chart
values are either positive or negative.

This chart type might also not be a good choice if the values of each data
series vary too much. The maximum number of data series (segments of a stacked
column) to be shown depends on the range of how much the values of each data
series vary: More than 5 data series will only work well in the case of little
variations.

Position the data series of central importance or interest directly on the axis
in order to best see its development over time.

**Grouped column charts**

![Figure EX 1.1-4: Grouped column charts](img/ex-1.1-4.png)

_Grouped column charts_ (short: grouped columns) show, in general, time series
for a primary scenario (e.g. AC or FC) in comparison with a reference scenario
(e.g. PY or PL). Two columns per category (_grouped columns_) represent these
two scenarios.

The columns of the primary scenario and the reference scenario overlap, the
reference scenario placed behind the primary scenario – either to the left or
right of the primary scenario (see bottom chart of the figure as well as the
paragraph on ”Scenario comparisons” in UN 4.1 “[Unify scenario
analyses](09-unify.md#un-41-unify-scenario-analyses)”). A third scenario could
be displayed using a _reference scenario triangle_. All other elements of a
grouped column chart are identical to single column charts.

Instead of using grouped columns, the primary scenario can be represented with a
single column with the reference scenario represented by reference scenario
triangles (see top chart of the figure).

**Horizontal pin charts**

![Figure EX 1.1-5: Horizontal pin charts](img/ex-1.1-5.png)

_Horizontal pin charts_ (short: horizontal pins) are used for the visualization
of relative variances in a time series analysis.

Horizontal pins consist of:

-   **Horizontal category axis:** see _single column chart_.

-   **Pins**: One _pin_ per time period or point of time extends from the
    category axis to the respective length. The pin has the size of a very thin
    column. Color the pin green or red corresponding with positive or negative
    relative variances respectively. The fill of the pinhead represents the
    primary scenario (see the paragraph on “Scenario comparisons” in UN 4.1
    “[Unify scenario analyses](09-unify.md#un-41-unify-scenario-analyses)”).
    Display the pin in the foreground, so that the length of the pin (see length
    X in the figure) is not hidden.

-   **Legends**: As there is only one data series, the legend (name of the data
    series) is part of the chart title.

-   **Data labels**: _Data labels_ name the values of the data series consistent
    with the length of the respective pins. Position the labels of positive
    values above the respective pins, labels of negative values below.

**Horizontal waterfall charts**

_Horizontal waterfall charts_ (short: _horizontal waterfalls_ or _column
waterfalls_) analyze root causes, over time, for the change or variance between
two or more statuses. Accordingly, horizontal waterfalls consist of two or more
_base columns and totals columns_. In between a base column and a totals column
there are _contribution columns_ demonstrating what led to the difference
between these two columns. The _contribution columns_ start at the end value,
i.e. the height, of the preceding column, and show the positive or negative
contribution as well as the accumulated contribution of all columns up to the
respective point of time.

There are two types of horizontal waterfalls:

![Figure EX 1.1-6: Growth waterfalls](img/ex-1.1-6.png)

**Growth waterfalls**: In _growth waterfalls_, base columns and totals columns
represent a stock measure (e.g. headcount, accounts receivable) at different
points in time (e.g. end of Q4 2012, 2013 and 2014). The contribution columns in
between represent the changes (increases and decreases) over time of this stock
measure.

(There is no vertical equivalent to the horizontal _growth waterfall_.)

![Figure EX 1.1-7: Horizontal variance waterfalls](img/ex-1.1-7.png)

**Horizontal variance waterfalls**: In _horizontal variance waterfalls_, base
columns and totals columns represent a flow measure (e.g. sales) at different
periods in time (e.g. 2015 and 2016) and/or regarding different scenarios (e.g.
PL and AC). The contribution columns in between represent the periodical
variances between the different time periods and/or scenarios.

The elements of a horizontal waterfall chart are the same as the elements of
single column charts. In addition, _assisting lines_ connect the end of a column
to the beginning of the succeeding column.

**Line charts**

![Figure EX 1.1-8: Line chart](img/ex-1.1-8.png)

In general, _line charts_ are used for the display of the temporal evolvement of
data series with many data points.

![Figure EX 1.1-9: Line chart with selective data labels](img/ex-1.1-9.png)

Many data points lead to small category widths. The advantage of line charts
over column charts is the simplified display of data (better _data-ink-ratio_).
In addition, they can better represent positive and negative values of more than
one data series than columns. On the other hand, lines tend to imply a
continuous timeline – practically non-existent in business communication.
Therefore lines should not be used for the presentation of data series with only
a few values.

Line charts cannot be “stacked” in order to show structure like in stacked
column charts. In the place of line charts for “stacked data”, _area charts_
offer a good solution (there is no layout concept for area charts in this
version of the guide yet).

Line charts with more than three intersecting lines tend to be confusing.
Instead, several smaller charts with one line each could be placed next to one
another (small multiples), particularly when the general trends of the lines are
to be analyzed – not the direct comparison of two data series (e.g. in comparing
seasonal developments of several years), see also EX 2.4 “[Replace spaghetti
charts](04-express.md#ex-24-replace-spaghetti-charts)”.

Line charts consist of:

-   **Horizontal category axis:** See _single column chart_. The semantic rules
    in part 3 suggest to use the category width (see width A in the
    first figure) for identifying the period type (see UN 3.3 “[Unify time
    periods](09-unify.md#un-33-unify-time-periods-and-points-of-time)”).

-   **Lines**: One or more _lines_ with _line markers_ represent the values of
    the respective data series. Use line thickness, line color, and line markers
    for meaning, see part “Semantic rules”.

-   **Legends**: _Legends_ label the data series. If the line chart shows only
    one data series, include the legend in the chart title. If the line chart
    shows two or more data series, the legend should be positioned to the right
    of the far right data point (left-aligned text, see first figure) or the
    left of the far left data point (right-aligned text, see second figure).
    Alternatively position legends close to the lines at any other place in the
    chart.

-   **Data labels**: _Data labels_ name the values of the respective data
    points. If possible, label maximum values (peaks) above the line markers and
    minimum values (valleys) below the line markers. In many practical
    applications it is not necessary to clutter the line chart by labeling every
    data point, see second figure on the left and SI 5.3 “[Avoid unnecessary
    labels](05-simplify.md#si-53-avoid-unnecessary-labels)”.

**Other horizontal charts**

Other chart types with horizontal category axes are *boxplot charts* (range
charts) and _area charts_. There is no specific notation concept for these chart
types yet however it can be easily derived from the notation concept of column
and line charts.

**Charts with vertical category axes**

Charts with vertical category axes (_vertical charts_) typically show structural
data. In general, present structural data of one period or one point of time in
the form of _bars_.

Use the vertical category as a structure axis. Horizontally, the visualization
elements represent the data per structure element (there is no need for a
horizontal value axis as the visualization elements carry their own values).
Structure axes run from top to bottom and show characteristics of structures
(e.g. products or countries). The sequence of these elements depends on the
intended analysis; see the UNIFY section about “Structure analyses”.

In general, the data series of a vertical chart is represented by (horizontal)
_bars_ (single, stacked, grouped), by _horizontal pins_, or by _waterfall bars_.
Do not use lines in vertical charts as they could be interpreted as trends or
developments, which do not exist in structure analyses.

_Horizontal pins_ can be considered very thin bars, but because of their
importance are dealt with in a separate section. A chart with horizontal pins is
called a _vertical pin chart_.

Here follows the grouping of _vertical chart types_:

**Single bar charts**

![Figure EX 1.1-10: Single bar charts](img/ex-1.1-10.png)

In general, _single bar charts_ (short: single bars) are used for the structural
analysis of one data series (e.g., products, countries, or divisions) for one
period or one point in time.

Single bar charts consist of:

-   **Vertical category axis:** The _vertical category axis_ with its labels
    represents the respective structure elements such as countries, products,
    etc. The category width (see width A in figure) should be the same for
    corresponding analyses.

-   **Bars**: One _bar_ per structure element extends from the category axis to
    the length representing the respective value. Display the bars in the
    foreground of the category axis, so that the length of the bar is not
    hidden. The part on “Semantic rules” suggests that the ratio of bar width to
    category width (see ratio B/A in figure) represents information about the
    measure type (see UN 3.1 “[Unify
    measures](09-unify.md#un-31-unify-measures)”).

-   **Legends**: As there is only one data series, the legend (name of the data
    series) is part of the chart title.

-   **Data labels**: _Data labels_ name the values of the data series consistent
    with the length of the respective bars. Position the labels of positive
    values at the right hand side of the respective bars, the labels of negative
    values at the left hand side.

**Stacked bar charts**

![Figure EX 1.1-11: Stacked bar charts](img/ex-1.1-11.png)

Stacked bar charts (short: stacked bars) represent more than one data series
(e.g., products, countries, or divisions) for one period or one point in time.

Stacked bar charts consist of:

-   Vertical category axis: See single bar charts.

-   **Bars**: The bars (see single bar charts) are divided into segments (Excel
    names them“data points”) representing the data series (stacked bars).

-   **Legends**: Legends (names of the data series) are positioned either above
    the top stacked bar or below the bottom stacked bar, with the bar segments
    defining their horizontal position: they are horizontally centered with the
    data labels of the respective bar segment. If a segment at the top or bottom
    is missing or has a very small size, its legend might need assisting lines.

-   **Data labels**: _Data labels_ name the values of the data series
    corresponding to the length of the respective bar segment. If the sum of the
    bar segments of a category is positive (bar pointing to the right), the
    label of the sum is positioned to the right hand side of the respective bar.
    If the sum is negative (bar pointing to the left), the label of the sum is
    positioned to the left hand side of the respective bar.

It must be pointed out that stacked bars should only be used if all chart values
are either positive or negative.

This chart type might also not be a good choice if the values of each data
series vary too much. The maximum number of data series (segments of a stacked
bars) to be shown depends on the range of how much the values of each data
series vary: More than 5 data series will only work well in the case of little
variations.

Position the data series of central interest directly at the axis in order to
best see its structure.

**Grouped bar charts**

![Figure EX 1.1-12: Grouped bar charts](img/ex-1.1-12.png)

In general, _grouped bar charts_ (short: grouped bars) show structure analyses
for a primary scenario (e.g. AC or FC) in comparison to a reference scenario
(e.g. PY or PL). Two bars per category (_grouped bars)_ represent _these_ two
scenarios.

The bars of the primary scenario and the reference scenario overlap, the
reference scenario placed behind the primary scenario – either above or below
(see the bottom chart of the figure as well as the paragraph on “Scenario
comparisons” in UN 4.1 “[Unify scenario
analyses](09-unify.md#un-41-unify-scenario-analyses)”).

A third scenario could be displayed using a _reference scenario triangle_. All
other elements of a grouped bar chart are identical to a single bar chart.

Alternatively, instead of grouped bars, the primary scenario can be represented
with a single bar and the reference scenario by reference scenario triangles
(see top chart of figure).

**Vertical pin charts**

![Figure EX 1.1-13: Vertical pin charts](img/ex-1.1-13.png)

_Vertical pin charts_ (short: vertical pins) are used for the visualization of
relative variances in a structure analysis.

Vertical pins consist of:

-   Vertical category axis: see _single bar chart_.

-   **Pins**: One _pin_ per structure element extends from the category axis to
    the respective length. The pin has the size of a very thin bar. It is
    colored green or red when representing positive or negative relative
    variances respectively. The fill of the pinhead represents the primary
    scenario (see the paragraph on “Scenario comparisons” in UN 4.1 “[Unify
    scenario analyses](09-unify.md#un-41-unify-scenario-analyses)”). Display
    pins in the foreground, so that the length of the pin (see length X in the
    figure) is not hidden.

-   **Legends**: As there is only one data series, the legend (name of the data
    series) is part of the chart title.

**Data labels**: _Data labels_ name the values of the data series corresponding
to the length of the respective pins. Position the labels of positive values at
the right hand side of the respective pins, the labels of negative values at the
left hand side.

**Vertical waterfall charts**

_Vertical waterfalls charts_ (in short: _vertical waterfalls_ or _bar
waterfalls_) analyze structural root causes for the difference between two or
more statuses. Accordingly, vertical waterfalls consist of two or more _base
bars_ and _totals bars_. In between a base bar and a totals bar there are
_contribution bars_ representing the contribution to the difference between
these two bars. Starting from the top base bar, _contribution bars_ always start
at the end of the preceding bar, showing positive or negative individual
contributions of the respective structure element as well as the accumulated
contribution resulting in the next totals bar.

There are two types of vertical waterfalls:

![Figure EX 1.1-14: Calculation waterfalls](img/ex-1.1-14.png)

**Calculation waterfalls**: In _calculation waterfalls_, base bars and totals
bars represent base and result measures (e.g. sales and EBIT) whereas the
contribution bars in between represent the additions and subtractions of other
measures (e.g. financial income and direct cost) in a calculation scheme. More
complex calculation schemes (e.g. profit and loss calculation) can have
intermediate subtotals bars.

(There is no horizontal correspondence to the vertical _calculation waterfall_.)

![Figure EX 1.1-15: Vertical variance waterfalls](img/ex-1.1-15.png)

**Vertical variance waterfalls**: In _vertical variance waterfalls_, base bars
and totals bars represent values at different periods or points in time (e.g.
January 1, 2013 and January 1, 2014) and/or different scenarios (e.g. PY and
AC). The contribution bars in between represent the variances in structure
between the different times and/or scenarios.

The elements of vertical waterfalls are the same as the elements of single bar
charts. In addition, _assisting lines_ connect the end of a bar with the
beginning of the succeeding bar.

**Remainder bar**

![Figure EX 1.1-16: Remainder bar](img/ex-1.1-16.png)

If a large number of elements need to be presented, then only the most important
elements can be displayed in one chart or on one page. In order to make the
analyses exhaustive, sort the elements by descending size, accumulating the
smallest elements, which cannot be depicted, in a _remainder bar_ (“rest of...”).
Separate the remainder bar from the other bars by a wider gap (see gap C in the
figure on the left).

Note: This remainder bar has to be excluded from some Structure analyses such as
averaging, ranking, and selecting.

**Other vertical charts**

Other chart types with vertical category axes are _vertical boxplot charts_
(range charts). There is no specific notation concept for this chart type yet
however it can be derived from the notation of the standard bar charts.

In general, do not use lines and areas in vertical charts as they might
underline a continuum of data non-existent in business communication.

**Charts with two values axes**

![Figure EX 1.1-17: Charts with two values axes](img/ex-1.1-17.png)

_Charts with two value axes_ show two-dimensional positioning of visualization
elements, which can provide new and interesting insights. *Scattergrams* arrange
points in a two-dimensional coordinate system.

![Figure EX 1.1-18: Bubble charts](img/ex-1.1-18.png)

*Bubble charts* (portfolio charts) have bubbles instead of points and use the
bubble area to show a third dimension. A fourth dimension could be presented via
pie segments within the bubbles (bubble pie charts).

Besides _scattergrams_ and _bubble charts_ there are other chart types with two
value axes, e.g. charts with horizontal axes representing a continuous timeline
(instead of fixed time categories) and charts with columns or bars of variable
width.

There are no specific notation rules for charts with two value axes yet. An
appropriate notation concept for these chart types can be derived from the
notation of column charts, bar charts and line charts with their data
visualization elements, legends, data labels, and axes.

## EX 1.2 Use appropriate table types

A *table* is a communication object in which data is arranged in two dimensions,
i.e. (vertical) _columns and_ (horizontal) _rows_. The _row header_ (row name)
describes the content of a row, the _column header_ (column name) the content of
a column. The data are positioned at the intersections of rows and columns
called _table cells_.

“One-dimensional tables” (tables with one or more columns but without row
headers) are called _lists_ and are not covered here.

_Table types_ are defined by a set of _columns_ and a set of _rows_ in order to
fulfill specific analytic and/or reporting purposes.

**Column types**

Column types are columns with similar content falling under similar column
headers. Typical column types are _time columns_ (with monthly or yearly data),
_scenario columns_ (with actual or plan data) and _variance columns_ (ΔPL or
ΔPY).

The following layout principles apply to all column types:

-   **Width**: Columns belonging to a certain column type should have an
    identical width. This column width should not depend on the text length of
    the respective column header.

-   **Orientation**: Right-align columns with numerical data. Left-align columns
    with non-numerical data (e.g. texts or product names). _Column headers_ have
    the same orientation as the rest of the column. Headers for combined columns
    can be centered or even left-aligned to increase legibility.

-   **Vertical lines and gaps**: Vertical lines separating different columns
    should be very light or even omitted. Use white vertical lines or white
    vertical gaps to mark the columns. In the following figures, different
    widths of these white lines resp. gaps are being used to separate and group
    columns. Separate columns of the same type by a narrow gap (see gap B1 in
    the figure in section “Scenario columns” et seq.). Use a wider gap to
    separate a group of similar columns from the next group (see gap B2 in the
    figure in section “Row header columns” et seq.).

Additional layout principals depend on the _column types_ described below.

**Row header columns**

![Figure EX 1.2-1: Row header columns](img/ex-1.2-1.png)

Row header columns contain the header texts of the rows. Often, these columns
are positioned at the very left of a table. In most cases, row header columns
are much wider than other column types.

Keep the texts of the row headers short by using abbreviations or footnotes in
order to omit too wide tables.

Use a wider gap (see width B2 in the figure) to separate the _row header column_
from columns with numbers.

**Scenario columns**

![Figure EX 1.2-2: Scenario columns](img/ex-1.2-2.png)

_Scenario columns_ show data for scenarios (e.g. previous year, plan, actual).
Use the same width for all scenario columns (depending on the number of digits).

For the sequence of scenario columns see UN 4.1 “[Unify scenario
ana­lyses](09-unify.md#un-41-unify-scenario-analyses)”.

**Variance columns**

![Figure EX 1.2-3: Variance columns](img/ex-1.2-3.png)

_Variance columns_ show data of absolute variances (e.g. ΔPL, ΔPY) or relative
variances (e.g. ΔPL%, ΔPY%).

**Time columns**

![Figure EX 1.2-4: Time columns](img/ex-1.2-4.png)

_Time columns_ show _time periods_ (for flow measures) or _points of time_ (for
stock measures).

Use a temporal order – from left to right – for the sequence of the time columns
(e.g. Jan, Feb, Mar, or 2013, 2014, 2015).

**Measure columns**

![Figure EX 1.2-5: Measure columns](img/ex-1.2-5.png)

_Measure columns_ show measures such as sales, headcount, or equity.

Displaying long measure names in column headers can be challenging. As the
column width should not depend on the length of the measure name, use the
abbreviations defined in the glossary instead.

Use a wider gap after intermediate results to expose the calculation scheme (see
width B2 in the figure on the left).

**Structure columns**

![Figure EX 1.2-6: Structure columns](img/ex-1.2-6.png)

_Structure columns_ show the elements of a structure dimension (e.g. countries
or products).

**“Thereof” columns**

![Figure EX 1.2-7: “Thereof” columns](img/ex-1.2-7.png)

If details of an aggregated column are shown in one or more column not totaling
to the aggregated column, these columns are called “thereof” columns.

The design of the _thereof columns_ should differ from other columns. E.g. use a
smaller font (see X in the figure) to expose a _thereof column_ and do not
separate it from the mother column (see columns _AL3_ and _AL3.1_ in the figure)
in order to show that it is part of it. A _thereof column_ is positioned at the
right hand side of the mother column.

**Remainder columns**

![Figure EX 1.2-8: Remainder columns](img/ex-1.2-8.png)

If the set to be presented in the columns has too many elements, accumulate the
less important or smaller elements in a _remainder column_ (e.g. 10 columns for
the top 10 countries and a remainder column titled“Rest of world” or “RoW”).

In the figure, the _remainder column_ “Other cost” has the same vertical gaps B1
as the other measure columns.

**“Percent of” columns**

![Figure EX 1.2-9: “Percent of” columns](img/ex-1.2-9.png)

Use “_Percent of_” columns to present important data of another column as shares
of a given total. A typical example for a “percent of” column is data of a
profit and loss statement as a percentage of sales.

“Percent of” columns have a smaller font size (see X) than the other columns.

**Totals columns**

![Figure EX 1.2-10: Totals columns](img/ex-1.2-10.png)

Position columns displaying _totals of a group of columns_ (e.g. quarters
totaling in years or products totaling in product groups) at the right hand side
of the columns belonging to this group. The design of the _totals columns_
should differ from other columns, e.g. highlighted by bold fonts or by light
gray background.

The column types described before refer to _single_ columns. The following
paragraphs present _combined_ columns i.e. _hierarchical_ and _nested_ columns.

**Hierarchical columns**

![Figure EX 1.2-11: Hierarchical columns](img/ex-1.2-11.png)

Hierarchies in dimensions may call for columns showing multiple levels. If
possible, the sibling elements belonging to the same parent element of a
dimension should be homogenous, mutually exclusive, and collectively exhaustive.

Separate parents by appropriate means, e.g. wider gaps. Display the parent
columns at the right hand side of their child columns (like _totals columns_).

In the figure, a wider gap B2 separates the two years (with four quarters each)
from each other.

**Nested columns**

![Figure EX 1.2-12: Nested columns](img/ex-1.2-12.png)

In _nested columns_, two column types are combined in such a way that the
columns of one type repeat iteratively within every column of the other type.
Separate the resulting groups of columns by appropriate means, e.g. wider gaps.

In the figure, wider gaps B2 separate the four years (with AC and PL data each)
from each other.

**Row types**

_Row types_ are rows with similar content falling under similar row headers.
Typical row types are _measure rows_ (e.g. sales, cost, profit) or _structure
rows_ (e.g. Italy, France, UK).

The following layout principles apply to all row types:

-   **Height**: Rows belonging to a row type should have an identical height
    (see height A in the figure in section “measure rows” et seq.).

-   **Horizontal lines**: Separating rows by light horizontal lines will
    increase the legibility.

Additional layout principals depend on the row types described below.

_Time periods and points of time_, _scenarios_, and _variances_ should be
displayed in columns rather than in rows.

**Column header rows**

![Figure EX 1.2-13: Column header rows](img/ex-1.2-13.png)

Column header rows contain the header texts of the columns. In most cases, these
rows are positioned at the very top of a table. In order to group columns two
and more column header rows might be necessary. If necessary, abbreviate column
header texts in order to fit in the preferred column width. Alternatively keep
column headers short by using footnotes.

In the figure the _column header row_ uses two lines in order to fit the column
header texts in the narrow columns.

**Measure rows**

![Figure EX 1.2-14: Measure rows](img/ex-1.2-14.png)

_Measure rows_ show measures such as sales, headcount, or equity.

Separate rows showing final or intermediate results of a calculation scheme
(_results rows_ _or totals rows_) by solid lines. Display results rows in bold
font or highlight them with light gray background. An additional gap B below a
results row will increase legibility.

**Structure rows**

![Figure EX 1.2-15: Structure rows](img/ex-1.2-15.png)

Structure rows show elements of a structure dimension (e.g. countries or
products).

**“Thereof” rows**

![Figure EX 1.2-16: “Thereof” rows](img/ex-1.2-16.png)

If details of an aggregated row are shown in one or more rows not totaling to
the aggregated row, these rows are called “thereof” rows. Place the aggregated
_above_ the “thereof” rows (in contrast to the _totals row_ positioned _below_
the rows of its group).

The design of the _thereof rows_ should differ from other rows. E.g. in the
figure, the _thereof row_ is of smaller height, written in a smaller font (see
X), not separated by a horizontal line, and has a right-aligned row header.

**Remainder rows**

![Figure EX 1.2-17: Remainder rows](img/ex-1.2-17.png)

If the structure dimension to be presented in the rows outline has too many
elements, accumulate the less important or smaller elements in a _remainder row_
(e.g. 7 rows for the top 7 countries and a remainder titled “Rest of world”).

Exclude remainder rows from some of the Structure analyses such as averaging,
ranking, and selecting.

In the figure, the _remainder row_ has the same row height A as the other
structure rows of this table example.

**“Percent of” rows**

![Figure EX 1.2-18: “Percent of” rows](img/ex-1.2-18.png)

Use “_Percent of_” rows to present important data of another row as shares of a
given total. A typical example for a “_percent of_” row is gross profit as a
percentage of sales.

“Percent of” rows have a smaller font size (see X) than the other rows.

**Totals rows**

![Figure EX 1.2-19: Totals rows](img/ex-1.2-19.png)

Place rows displaying _totals of a group of rows_ (e.g. countries totaling in
regions or products totaling in product groups) below the rows of this group and
separated them by solid lines.

The design of the _totals rows_ should differ from other rows, e.g. highlighted
by bold fonts or by light gray background.

The row types described before refer to _single_ rows. The following paragraphs
present _combined_ rows i.e. _hierarchical_ and _nested_ rows.

**Hierarchical rows**

![Figure EX 1.2-20: Hierarchical rows](img/ex-1.2-20.png)

Hierarchies in dimensions may call for rows showing multiple levels. If
possible, the sibling elements belonging to the same parent element of a
dimension should be homogenous, mutually exclusive, and collectively exhaustive.

Separate parents by appropriate means, e.g. wider gaps (see additional gap B in
the figure). Display the parent rows _below_ their child rows (like _totals
rows_).

**Nested rows**

![Figure EX 1.2-21: Nested rows](img/ex-1.2-21.png)

In _nested rows_, two types of rows are combined in such a way that the rows of
one type repeat iteratively within every row of the other row type.

Separate the resulting groups of rows by appropriate means, e.g. wider gaps (see
additional gap B in the figure).

**Table types**

![Figure EX 1.2: Table types](img/ex-1.2.png)

Table types are distinguished by their analytic purpose in time series tables,
variance tables and cross tables. Tables serving more than one analytic purpose
are called combined tables.

**Time series tables**

![Figure EX 1.2-22: Time series tables](img/ex-1.2-22.png)

_Time series tables_ are used for time series analyses, combining time columns
with measure rows or structure rows.

A typical example for a _time series table_ is a sales analysis by countries
(rows) and years (columns).

**Variance tables**

![Figure EX 1.2-23: Variance tables](img/ex-1.2-23.png)

_Variance tables_ are used for scenario analyses, combining scenario columns and
variance columns with measure rows or structure rows.

A typical example for a _variance table_ is a sales analysis by countries (rows)
showing different scenarios and different variances (columns).

**Cross tables**

![Figure EX 1.2-24: Cross tables](img/ex-1.2-24.png)

_Cross tables_ are used for Structure analyses, combining structure columns with
structure rows.

A typical example of a _cross table_ is a sales table with countries in rows and
products in columns.

**Combined tables**

![Figure EX 1.2-25: Combined table 1](img/ex-1.2-25.png)

_Combined tables_ are used for multiple analyses. A combined table uses more
than one _column type_ and/or more than one _row type_ presented either side by
side or nested.

The first figure shows a hierarchical structure of countries on three levels in
the rows. The columns are nested: scenarios and variances are the same for both
time periods _November_ and _January_November_.

![Figure EX 1.2-26: Combined table 2](img/ex-1.2-26.png)

The second figure shows the measures of a calculation scheme in the rows. The
columns are nested: The four quarters and the annual total are the same for both
years.

![Figure EX 1.2-27: Combined table 3](img/ex-1.2-27.png)

The third figure shows the same rows as the second one (measures of a
calculation scheme). The nested columns now show PY and AC data as well as
absolute and relative variances for two markets.

## EX 2 Replace inappropriate chart types

Inappropriate charts make it hard to perceive the message. Knowing the correct
usage of chart types helps in replacing inappropriate visualizations, such as
pie charts, speedometer visualizations, radar charts, and spaghetti charts, with
those chart types better suited.

## EX 2.1 Replace pie and ring charts

![Figure EX 2.1: Replace pie and ring charts](img/ex-2.1.png)

_Pie_ and _ring charts_ are [circular
charts](http://en.wikipedia.org/wiki/Circle) dividing some total into
[sectors](http://en.wikipedia.org/wiki/Circular_sector) of relative proportion,
but there are better ways to illustrate the numericalproportions of segments,
e.g. bar charts or charts with stacked columns, see Figure EX 2.1.

_Pie charts_ allow for one-dimensional analyses only, and therefore seldom
convey revealing insights. However, some useful applications for pie charts
exist, for example when market sizes and/or market shares for one period need to
be allocated to certain regions on a map (see CH 3.3 “[Avoid misleading colored
areas in maps](07-check.md#ch-33-avoid-misleading-colored-areas-in-maps)”). As
opposed to column or bar charts, pie charts can be positioned on a specific
point on a map.

## EX 2.2 Replace gauges, speedometers

![Figure EX 2.2: Replace gauges, speedometers](img/ex-2.2.png)

Often found as part of a so-called dashboard, _speedometers_ are probably one of
the most useless visualizations out there. They take up way too much space and
have often confusing color coding and scaling. In general, bar charts showing
the respective structures or columns charts showing the respective development
over time are better choices, see Figure EX 2.2.

## EX 2.3 Replace radar and funnel charts

![Figure EX 2.3: Replace radar and funnel charts](img/ex-2.3.png)

So-called _radar charts_ (also called _net charts_ or _spider charts_) are
frequently used for evaluating purposes. Having no advantage over bar charts and
having, actually, many weaknesses, use them only for two-dimensional analyses
(e.g. comparing young-old with rich-poor). Willard C. Brinton wrote almost 100
years ago: “This type of chart should be banished to the scrap heap. Charts on
rectangular ruling are easier to draw and easier to understand.”

Of course, if the circular arrangement has meaning (such as the compass
direction), this kind of chart can be very valuable, but these types of analysis
are not typical in business reporting.

_Funnel charts_ are misleading when the size of the area displayed does not
correspond to the respective numerical values – an issue applying also to other
artificial chart forms (e.g. spheres) in which the length, area, or volume do
not correspond to the numerical values.

## EX 2.4 Replace spaghetti charts

![Figure EX 2.4: Replace spaghetti charts](img/ex-2.4.png)

A chart with more than three or four intersecting lines (“spaghetti chart”) can
be more confusing than several smaller charts with one line each placed next to
one another (small multiples), particularly when evaluating the shape or the
trend of the lines, see Figure EX 2.4.

However, when needing to compare exactly the height of data points of several
lines, spaghetti charts cannot be avoided.

## EX 2.5 Replace traffic lights

![Figure EX 2.5: Replace traffic lights](img/ex-2.5.png)

“Traffic lights” with green, red, and yellow areas are a popular form of
visualization but contain little information per area used. However, they can be
used for analyses showing “yes or no” decisions or situations similar to real
traffic lights. In all other cases replace them with more suitable means of
(analog) representation such as bar charts, see Figure EX 2.5.

## EX 3 Replace inappropriate representations

From a perceptual perspective, avoid all visual representations requiring time
consuming analyses or additional explanations, particularly the popular use of
merely conceptual representations and all forms of texts, including bullet
lists.

## EX 3.1 Prefer quantitative representations

![Figure EX 3.1: Prefer quantitative representations](img/ex-3.1.png)

Due to the time constraints usually involved with presentations, conceptual
graphs prove less suitable than charts, photos, maps, etc. For a one-hour
presentation, do not use more than three or four conceptual representations. Do
this only if they are absolutely essential for comprehension. The audience will
understand charts and pictures (photos, drawings, etc.) better and faster,
see Figure EX 3.1.

## EX 3.2 Avoid text slides in presentations

![Figure EX 3.2: Avoid text slides in presentations](img/ex-3.2.png)

Avoid all forms of text slides in presentations. Texts should either be recited
or written in a handout. A few exceptions to this rule are specific texts being
discussed such as definitions, quotes, etc. In general, all forms of lists
(bullet points) should appear only in the written handout, not projected on the
screen. True, if someone sees and hears something simultaneously, he remembers
it better than when he just hears it, but bear in mind texts are not considered
something merely to be seen – they must be read and understood, see Figure EX
3.2.

## EX 4 Add comparisons

Visual perception is strongly based on setting one perceived object in relation
to another. Adding meaningful comparisons helps the eye evaluate faster, the
main purpose of charts.

## EX 4.1 Add scenarios

![Figure EX 4.1: Add scenarios](img/ex-4.1.png)

Scenarios such as “plan” and “previous year” are the most common references for
comparison purposes. Add them whenever available. Use a standardized scenario
notation for faster comprehension, see Figure EX 4.1.

## EX 4.2 Add variances

![Figure EX 4.2: Add variances](img/ex-4.2.png)

Having added scenarios for comparison purposes, the visualization of variances
makes it easier to evaluate the situation. Use a standardized notation of
variances for faster comprehension, see Figure EX 4.2.

## EX 5 Explain causes

Present data more understandable by showing interrelations, i.e. causes and
dependencies. Seeing the entire context, especially extreme values and deviant
values, helps to explain causes. Details increase not only the level of
credibility but also comprehension. Use charts to prove, explain, and render
something plausible, not to serve merely as decoration. This section focuses on
the explanation of causes by using tree structures, clusters, and correlations.
A more structured approach to increasing information density is discussed in the
chapter “CONDENSE – Increase information density”.

## EX 5.1 Show tree structures

![Figure EX 5.1: Show tree structures](img/ex-5.1.png)

The presentation of the assumptions or basic data upon which a business analysis
is based, results not only in better understanding, but also makes it more
convincing. A good choice is the display of calculated measures in a tree
structure, see Figure EX 5.1 (see also CO 5.2 “[Show related charts on one
page](06-condense.md#co-52-show-related-charts-on-one-page)”).

## EX 5.2 Show clusters

![Figure EX 5.2: Show clusters](img/ex-5.2.png)

With the help of clusters in two-dimensional and three-dimensional forms, large
amounts of data very often can provide interesting and new insights, see Figure
EX 5.2.

## EX 5.3 Show correlations

![Figure EX 5.3: Show correlations](img/ex-5.3.png)

When comparing several data series, correlations are often sought in order to
better understand the interrelations. Suitable rankings and comparisons can
facilitate the understanding of patterns, see Figure EX 5.3.

[← Organize content](02-structure.md) | [Avoid Clutter →](05-simplify.md)