Assume your IP3 is +20dBm. The 3rd order distortion drops 30dB, per 10dB input power reduction. Thus -10dBm input RF power drops the distortion by 3*(+20dBm - (-10dB)) or by 90dB below the IP3, or to +20- (90) to -70dBm;
[ note my Initial Error in computing the distortion level: had gone 90dB below +10dBm and that is wrong; should be 90dB below the IP3, which is -70dBm distortion products power; another (correct) way to compute would be to take 2/3 of the drop in distortion (or 2/3of 90dB) and refer that reduced level to the RF input; that would give 60dB below RF input of -10dBm, or the -70dBm power level for distortion products.]
Your input RF is -10dBm; the distortion is -70dBm, your Harmonic Distortion ratio is 60dB.
Is that satisfactory for your signal chain?
However, the two input amplifiers may overdrive your mixer.
You need to sketch out the gain chain and the IP3, and compute the resultant 3rd order.
What input power level to expect from the antenna? Suppose you need a +10dB SNR, for 10MHz bandwidth. Suppose your front end antenna matching, coax, and LNA produce a 4dB degradation of SNR. Just add these up, and reference to the Boltzmann Thermal Noise in 1Hertz bandwidth at +17 degree Centigrade:
+70dB for the increased noise power of a 10,000,000Hz bandwidth
+10dB SNR for acceptable bit errors, packet errors and re-transmission requests, and trivial Bit_Error_Correction protocols
-174 + 4 + 70 +10 = -174 + 84 = -90dBm.
Since 0dBm across 50 ohm is 0.223 volts RMS and 0.632 vols PeakPeak, and -120 dBm is 0.223 microVolts RMS and 0.632 microVolts PeakPeak, your power level of -90 dB is 30dB more power (1,000X) and is sqrt(1,000) or 31.6 higher voltage; thus your input RMS power (needed for LNA thinking) is -90dBm and 0.223 * 31.6 or about 7 microVolts RMS.
Do you understand this diagram?
simulate this circuit – Schematic created using CircuitLab