Good designers spend a great deal of time sweating over typography. They agonise over typefaces, iterate through type scales and meticulously apply white space, all in the service of the reader. Then comes along a table with the temptation to get creative, and all thoughts of the reader go out of the window. And yet tables are there to be read, referenced and used, not merely looked at.
Set tables as text to be read
Tables come in many forms. Some contain simple numbers, others are complex with mixtures of numeric data and textual information. Some require reading row by row, others are scanned vertically. The potential use for tables is as varied as the written word. They can be financial statements, bus timetables, multilanguage dictionaries, tables of contents, numerical conversions, pricing options, feature comparisons, technical specifications, and so on.
Despite the huge variation in table size, complexity, contents and purpose, every table shares two simple design principles: they should be readable and support a sense of the data held within. They should not be prettied up to satisfy a sense of aesthetic when simply looked at. That said, a well-designed table can still be a thing of beauty but with the form following the function. Tables are not pictures of data: they are catalogues of data to be perused, parsed, referenced and interrogated. A well-designed table will enable the information to be read and understood, and will reveal the patterns and correlations inherent in the data. As Jan Tschichold, the virtuoso of typography design, put it in Asymmetric Typography1:
Tabular matter need no longer be a rather unpleasant job to design: on the contrary, it can become a really charming and artistic exercise, in no way less interesting than any other area.
Wherever possible plan the readability of every table in advance. Your design process should be an investigation into making the data undemanding to read, simple to follow and easy to extract.
Just as you wouldn’t design body text with the aim of fitting as many words as possible on the screen, so you shouldn’t treat designing a table as an exercise in cramming as much data as possible into one space. You might be tempted to reduce the text size of your table – and if the data is entirely numeric you might be able to get away with it. Your reader should still be able to be comfortably read and interpret the table from their normal position, without needing to lean in.
DON’T STRETCH TABLES
Many designers will instinctively apply a width to their tables – just as they might an image – stretching them to fill the text column or page. And that is the appeal of setting tables full-width: you can make them look somewhat image-like when viewed from afar. However, while a table spread across the screen might look preferable from a distance, on closer inspection it will be harder to read as the data will be unnecessarily separated. To add insult to injury, tables set full-width are often replete with background colours and borders to give the table further the texture of an image, when what your reader really requires is the texture of text. For the sake of your readers, avoid these temptations.
You might consider making all the columns an even width. This too does nothing for the readability of the contents. Some table cells will be too wide, leaving the data lost and detached from its neighbours. Other table cells will be too narrow, cramping the data uncomfortably. Table columns should be sized according to the data they contain. Columns of small numbers should be narrow, and columns of paragraphs should be relatively wide. This sounds like a lot of effort, and for a print designer it would be, as they would have to size each column manually in their layout software. Fortunately, web browsers are very clever when it comes to laying out tables and will do all that hard work for you. Browsers have been laying out tables automatically according to complex algorithms since long before CSS came along – just let them do their thing.
KEEP TABLE FURNITURE AND FILLS TO A MINIMUM
The statistician and information designer Edward Tufte introduced the concept of data-ink in his 1983 classic, The Visual Display of Quantitative Information. He defines data-ink as ‘the non-erasable core of the graphic’, whereas non-data-ink is the ink used in the graphic, not to directly represent data but for scales, labels, fills and edges. Tufte goes on to define the data-ink ratio as the proportion of ink that is used to present actual data compared to the total amount of ink used in the entire graphic. The goal is to design a graphic with the highest possible data-ink ratio (tending towards 1.0) without eliminating what is necessary for effective communication.
Where Tufte talks about graphics he includes charts, diagrams and tables, and where he uses ‘ink’ we can think of pixels. In terms of tables, he’s saying that we should remove almost everything in the design which is not data or white space. Minimise furniture, maximise information. This is an ideal first principle to bear in mind when considering the typographic design of a table.
As a starting point, avoid any border or frame surrounding the table. This is a Victorian embellishment which is entirely unnecessary as text alignment will shape the table just fine.
Try to achieve a readable table using just alignment, spacing and grouping. Avoid zebra striping, tints and fills, and any other backgrounds. These can be superficially pretty but are usually a distraction. They serve to distort the meaning of the data by highlighting every other row to the detriment of neighbouring rows. Only use tints as a subtle means of guiding your reader’s eyes, and then only if you cannot arrange the data to that end. If you choose to tint, do so only in the primary direction of reading: down if lists, across otherwise.
When it comes to lines and borders between rows and columns – typographically referred to as rules – the same applies: use them judiciously and preferably not at all. In Asymmetric Typography Jan Tschichold sums this up wonderfully:
Tables should not be set to look like nets with every number enclosed. Try to do without rules altogether. They should be used only when they are absolutely necessary. Vertical rules are needed only when the space between columns is so narrow that mistakes will occur in reading without rules. Tables without vertical rules look better. Thin rules are better than thick ones.
Avoid using row or column borders unless the data alignment, spacing and grouping are not sufficient to guide your reader’s eye. If you do need to use rules for this purpose, use them in one direction only and employ a lighter colour to reduce the impact of the lines: you are making a distinction, not constructing a barricade.
LEFT-ALIGN TEXT, RIGHT-ALIGN NUMBERS, AND ALIGN HEADINGS WITH DATA
In the spirit of treating tables as artefacts to be read, don’t centre text within tables. Align table text as you would anywhere else; that is, aligned left. As text in tables tends to end up in narrow columns, don’t justify the text either – leave it ragged-right – or you will end up with rivers flowing down the tables, potentially causing confusion and certainly harming readability. You can hyphenate, however, particularly if the table columns would otherwise have a pronounced rag.
Right-align numbers to help your reader make easier comparisons of magnitude when scanning down columns. To aid scanning in this manner you will need consistent precision of your numeric data; that is, use the same number of decimal places.
For consistency and ease of understanding, match the alignment of headings to the alignment of the data. Right-align headings of numeric data and left-align headings of columns with text, for example: