By far the most common genre of plot, bar (or column plots) probably outnumber all other types by a factor of three. Most of these plots are workmanlike and unmemorable. Some of them, however, utilize techniques that designers of plots should consider for their arsenal.
We start with Davies Ward Barometer 2007 [pg. 29] because of its conventional treatment of the bar plot. The labels for each bar are on the left axis, the bars are sorted from high to low (top to bottom), and the percentages are given at the right end of each bar. Some of the elements of this plot that are not standard, however, include the black outline of each bar, the tick marks at the base that look like staples around the label, and the choice to cut off responses under 4%.
The next plot, from Freshfields Bruckhaus Crisis 2013 [pg. 10], centers each segmented bar at zero, a neutral position. It also uses colors for the segments and embedded percentages to convey the data collected by the survey. Note, too, that the plot has no y-axis line.
KL Gates GCDisruption 2017 [pg. 6] chose to represent negative values, decreasing percentages, as a different color bar (purple), located in the lower right, with a space between it and the positive values. This technique is confusing. It would have been better to present the negatives on the left side of zero, as the preceding Freshfields Bruckhaus plot does.
DLA Piper RE 2017 [pg. 8] introduces several techniques. First, the designer places the labels of the bars within the bars, sometimes even stretching them outside to the right of the bar. Second, the firm summarizes the key point of the plot in the first bullet on the right and adds more context in the second bullet. Third, the explanation in the second bullet has a footnote regarding its source. Fourth, unlike most bar plots that put the percentage or number at the end of the bar, this one stacks them flush right.