Sed has the ability to define regions in the regular expression definition area. This regions can be then back referenced using the special characters "\1" to "\9". To define a region it has to be surrounded with parentheses.
As an example, lets say you have a nine characters long line that has to be splitted into three three-character regions:
Note I'm passing the "-r" argument to sed, otherwise I would have to escape the parenthesis characters to prevent sed to take them as literals.
Following is a "real life" example. I have a plain text file with email addresses and contact names with this format:
name_1:last_name_1:mail_1@domain.com
name_2:last_name_2:mail_2@domain.com
name_3:last_name_3:mail_3@domain.com
name_4:last_name_4:mail_4@domain.com
This could be difficult to read as the file grows. With just one line using sed defining regions, the output of a cat command can be formatted in a "human readable" format: