Rasputin - 8:21 am on May 25, 2012 (gmt 0)

I would say that our 'penguin' sites recovered a few % around the 19th May while non-penguin sites fell a similar amount at the same time (no changes made - they are mostly old, small sites), but the variation after a few days is small enough that it could possibly just be seasonal variation.

It is also possible that the introduction of Knowledge Graph is distorting the figures for our (travel) sites that fell a small amount around the same time (or the sunny weather in the UK could have got people away from their computers...)