The accuracy of previous measurements could be affected by stray light coming from the high power pump laser.

I carefully checked the photo-detector ad I found two small filters inside a threaded pipe attached in front of the detector. They are supposed to suppress the stray light signal.

Those filters are Heat Absorbing Glass: 

 - 3 mm-thick KG-3 glass. It filters out 1064 nm http://www.edmundoptics.com/document/download/352659

 - 2.5 mm-thick R-60 glass. It is the red-color filter used to block most of daylight. http://www.edmundoptics.com/optics/optical-filters/longpass-edge-filters/longpass-glass-color-filters/66043/?print=Pdf

In order to check if the pump stray light is well stopped by the filters, I switched off the probe (to avoid to have any true absorption signal, in purpose) and see if there is a different signal with and without the pump laser.

Since switching the probe OFF makes the DC signal almost 0, I don't divide the AC signal by the DC.

I took 1h of data with the pump ON; and 1h of data with the pump OFF. I attach the plot of the raw AC signal (X and Y) in the two cases, red and blue clouds of points.

The standard deviations of the X and Y signals are 1microV in both clouds. The difference between the means of the two clouds is 0.2microV.

Comparing with the order of 50microV of last absorption measurements, I conclude that pump stray light don't contribute to the signal.