@JoostWesseling if I look at the time series of your v1 Egg, it is true that most of the time it reports zero. But occasionally it does report non-zero values (at least for CO).
It's always been a problem with the v1 Egg that they are not calibrated and we do our best to make them report reasonable values before shipping, but their 'interpreted' response in context often has this problem of the settings not generalizing at a customer location. It's one of the main things that drove us to fundamentally different (electro-chemical rather than MOS) sensor technology in the v2 Egg.
There are a couple of things you can do. First, the "raw" values give you a clearer idea of what's going on in a relative sense. Have a look at your NO2 readings for example you can see that the raw measurements are varying even though the interpreted response is always zero. The reason is that the interpretation would yield a negative concentration, and we truncate that to zero in software.
.
If you are feeling more ambitious, you can tinker with the calibration setting using an Arduino sketch I wrote a while back called AQE_Calibrate. The whole point of this sketch is to let you set what R0 (which is the characteristic resistance of a given sensor). What we do is we let the Egg run for a while in our lab, and then type in the average resistance reported, and 0.15 for the R/R0 reading (which I believe corresponds to around 20ppb according to the datasheet). You can do the same sort of thing with AQE_Calibrate, but use your timeline data from Xively to set the resistance (i.e. skip the data collection period and just type in the values).