Antenna cross polarization size and wifi

Why cross-polarization ratio is important for lineary polarized antennas?

The question reason is one test. I have two wifi 802.11n MIMO modules and did throughput test two times. At first time AccessPoint Tx0 signal goes through attn and cable only to STAtion Tx0, AP Tx1 -> STA Tx1. At second time AP Tx0 -> attn -> splitter + two cables to STA Tx0 and STA Tx1; AP Tx1 -> attn -> splitter + two cables to STA Tx0, Tx1.
Throughputs are same at booth test-setups.

I also did test on 10km link - I change AP antenna (linear polarization) with ~30dB cross-polarization ratio to antenna with ~15dB cross-polarization ratio and throughput (155Mbps) doesn't change. Datarate is 300Mbps (40MHz, 2x2, MCS15), but real throughput is only ~155Mbps - it's maximal throughput of those wifi modules.
So, seems that for MIMO wifi antenna cross-polarisation ratio isn't important. Where I'm wrong?

The cross-polarization ratio in antennas is used to share two separate channels on one frequency.
If you only operate one data link on one frequency, your observation would not reveal anything.
Only if you need to transmit two data links on one frequency, from one location to another, you can use two antennas on each end, and e.g. one transmitting vertically polarized wave will send its data signal to the vertically polarized receiving antenna, and vice versa.
If your antennas will have a good cross-pol specification each, then the data link with the horizontally polarized antennas will not interfere with the vertically polarized antennas in the other link.
The use of the orthogonally polarized antennas require fixed pointing of antennas used. Systems like this are used for satellite links and for microwave terrestrial links. Changes in pointing or involving reflections along the path degrade the cross-pol ratio. As Wifi is typically used in a varying environment, I do not think the use of orthogonally polarized antennas could bring an advantage.