In recent years, various Blind Face Restoration (BFR) techniques were developed. These techniques transform low quality faces suffering from multiple degradations to more realistic and natural face images with high perceptual quality. However, it is crucial for the task of face verification to not only enhance the perceptual quality of the low quality images but also to improve the biometric-utility face quality metrics. Furthermore, preserving the valuable identity information is of great importance. In this paper, we investigate the impact of applying three state-of-the-art blind face restoration techniques namely, GFP-GAN, GPEN and SGPN on the performance of face verification system under very challenging environment characterized by very low quality images. Extensive experimental results on the recently proposed cross-quality LFW database using three state-of-the-art deep face recognition models demonstrate the effectiveness of GFP-GAN in boosting significantly the face verification accuracy.