Convolution normalisation factor

Where does the 0.125f factor come from when the impulse is normalised?

static float calculateNormalisationFactor (float sumSquaredMagnitude)
    if (sumSquaredMagnitude < 1e-8f)
        return 1.0f;

    return 0.125f / std::sqrt (sumSquaredMagnitude);

I’m sure there is a good dsp reason behind it, but I’m curious because the audio output is quite weak here, so I was just wondering if that could come from there.