Herivel realised that lazy operators would move the rotors by only a few letters – or not even move them at all – and that the indicators would all cluster around the original settings: "If the intercept sites could send us the indicators of all the first messages of the day for the individual German operators, there was a sporting chance that they would cluster around the ring settings for the day."
That would allow the codebreakers to narrow down the possibilities from 17,576 to a couple of dozen settings, which they could test out individually by hand. "The next day I went back to Hut 6 in a very excited state and told my colleagues of this idea. 'Oh, brilliant,' they all said."
It wasn't laziness, actually, but stress and overwork that did it:
Welchman had it tested out daily, but it did not work until two months later, in May 1940, as the Germans invaded France and the operators came under pressure of combat. Then the Herivel Tip worked – "It was a very exciting moment," Herivel later recalled.
...in typical British understatement.
The thing I love about this is that Herivel was recruited as a mathemetician, but had his insight because he understood what people would do. Numbers do not manipulate themselves.