Wide tables that should be long

• The layout is natural for reading and for typing: one row per thing, one column per period.
• A new column was added each month, since that was easier than adding rows.
• The table is the output of a pivot, saved as if it were the underlying data.

• Filtering or grouping by the spread-out variable is not possible, because it is not a column.
• Every new period changes the shape of the table, so anything built on it has to be edited.
• Combining files fails when each has a different set of period columns.

• Look for headers that are values instead of names: dates, years, or categories.
• Look for a run of columns that all hold the same kind of measurement.
• Ask whether next month's data would add rows or columns. If columns, the table is wide.

• Identify the columns that identify each row, and the columns that are really one variable spread out.
• Reshape so each row holds one identifier, one value of the spread-out variable, and one measurement.
• Name the two new columns for what they hold, and tidy the former headers into proper values, such as full dates.
• Keep the wide layout for presentation if people prefer it. Wide is a fine way to show data, just a poor way to store it.

• Merged and multi-row headersThe column names are spread over two or more rows, often with merged cells, so no single row names every column.
• One table split across sheetsRows that belong in one table are spread over several sheets or files, one per month, region, or person.
• Multiple values in one cellA single cell holds a list, or two different facts, joined by a comma, slash, or line break.

TermsWide format, Long format, Pivot, Tidy data