[nkron]

Here's my favorite example of that:

http://blog.chriszacharias.com/page-weight-matters

YouTube increased the efficiency of their site which ended up increasing average page load time - more people across the globe could now use the site when before it was too slow to bother with.

[aaron695]

250KB -> 100KB

Youtube use would have to be > 2.5 times as big after for the resource usage to have actually grown.

Not Jevons paradox's, not sure what to call it.

[TheCoelacanth]

Not really, because users with a high page load time can skew the average by a very large amount.

Hypothetically, let's say that previously they had 10 users who all took 2.5 seconds to load the page, so their average page load time is 2.5 seconds. After the change, those users all drop to a 1 second page load time. They also attracted one new user who now only has to wait 20 seconds instead of 50 seconds. Their average page load time is now 2.7 seconds but their usage only increased by 10%.

[shoo]

or, since we're talking about an average, not median, they attracted one new user who is now prepared to wait a very long time