While I was employed at Origin Systems it was very common to send games into test way before they were ready. The games were broken, had few features, and testing was nothing more than a waste of time. Through sheer force of will, Origin's product development and testing teams pulled some amazing product to completion in spite of this mistake. It's equally bad to send a game into test at the last minute, or without a reasonable plan.
By the time a game is content complete, you can't really add significant development resources to shorten the testing cycle. You can add testers like crazy, but the development team won't be able to keep up. Face it, if a programmer wrote some buggy code it might take ten times as long for another programmer to jump in and figure things out.
Every game has inertia to it, and before the game is shipped the team will have fixed a finite number of bugs, which equates to a very definable amount of work. When you hit the sweet spot, and your game goes into test at the right time, you'll have more fun at the end. More importantly, you'll make a better game.