FiveThirtyEight has an article about how the LibDems normally need two elections to win a seat:
[T]hey tend to win seats not in a single election cycle, but after first reaching a critical threshold of support an election before. While still losing to the major party in the constituency, this baseline of support provides the “plausibility” factor that can turn tactical and ideological voters to the party.
In 2005, seats that the Lib Dem picked up showed a particular voting trend from 1997 to 2005. In nearly every case, the 2001 election saw a swing to the Lib Dems, usually pulling votes from both Labour and the Tories. And, the stronger that swing was in 2001, the bigger the swing they experienced in 2005. Similarly, many of the seats in which they made progress but did not win in 2005 are now key pickup opportunities for the Lib Dems in 2010.
I did some quick calculations to see how much this matters.
At the moment, the LibDems notionally hold 76 seats and are in second place in another 194 seats.
If there is a uniform 5% swing from Labour to LibDem, the latter will get 118 seats, but they will suddenly be in second place in 332 seats!
In other words, although a LibDem vote can seem wasted in many constituencies, it might just prepare the ground for a LibDem landslide in 2014 or 2015.