The so-called unknown absorber in the clouds of Venus is an important absorber of solar energy, but its vertical distribution remains poorly quantified. We analyze the 283 and 365-nm phase curves of the disk-integrated albedo measured by Akatsuki. Based on our models, we find that the unknown absorber can exist either well-mixed over the entire upper cloud or within a thin layer. The necessary condition to explain the 365-nm phase curve is that the unknown absorber must absorb efficiently within the cloud scale height immediately below the cloud top. Using this constraint, we attempt to extract the SO$_2$ abundance from the 283-nm phase curve. However we cannot disentangle the absorption by SO$_2$ and by the unknown absorber. Considering previous SO$_2$ abundance measurements at mid-infrared wavelengths, the required absorption coefficient of the unknown absorber at 283~nm must be more than twice that at 365~nm.