It is essentially the only metric available that is readily comparable between countries. Also, whatever definition you use for "real unemployment rate" that rate will be linearly dependent on the U-3 metric. If the U-3 number goes up, then "real unemployment" goes up and likewise if it goes down.

Those who are interested can read the report to the 13th International Conference of Labour Statisticians (yes, there are conferences for everything) in which the unemployment definition which U-3 is based on was adpoted: http://www.ilo.org/public/libdoc/ilo/1982/82B09_438_engl.pdf