Conventional nanoindentation testing generally uses a peak penetration depth of less than 10 % of thin-film thickness in order to measure film-only mechanical properties, without considering the critical depth for a given thin film-substrate system. The uncertainties in this testing condition make hardness measurement more difficult. We propose a new way to determine the critical relative depth for general thin-film/substrate systems; an impression volume analyzed from the remnant indent image is used here as a new parameter. Nanoindents made on soft Cu and Au thin films with various indentation loads were observed by atomic force microscope. The impression volume calculated from 3D remnant image was normalized by the indenter penetration volume. This indent volume ratio varied only slightly in the shallow regime but decreased significantly when the indenter penetration depth exceeded the targeted critical relative depth. Thus, we determined the critical relative depth by empirically fitting the trend of the indent volume ratio and determining the inflection point. The critical relative depths for Cu and Au films were determined as 0.170 and 0.173, respectively, values smaller than 0.249 and 0.183 determined from the hardness variation of the two thin films. Hence the proposed indent volume ratio is highly sensitive to the substrate constraint, and stricter control of the penetration depth is needed to measure film-only mechanical properties.