When you run a basic sed command, it will autoprint the pattern space (a line of input) once it is done running the script against it. That means every line will get sent to stdout.

You can supress the autoprint functionality with the -n flag like so:

$ seq 100 | sed -n 's/1$/one/'

You can then add the p flag to the end of the substitute command to tell it to print any line that was affected by that substitution after the substitution has been applied.

$ seq 100 | sed -n 's/1$/one/p'
one
1one
2one
3one
4one
5one
6one
7one
8one
9one

For all numbers between 1 and 100, this matches those that end in 1 and substitutes that 1 for one. And then it is only those lines that go to stdout.

If you used the p flag without -n, every line would autoprint and then you'd get duplicate output for each line that had a substitution.

See man sed for more details.