High-precision single-receiver user positioning requires the combination of GNSS measurements with externally-provided random corrections to ensure unbiased estimation of an ambiguity-resolved user position. While the unbiasedness property has been shown to remain intact in the computed PPP-RTK parameter solutions, the same cannot be said for the minimum-variance optimality property. In this contribution we study and present the pivotal role of properly specifying the weight matrix for achieving best linear unbiased estimation of the single-epoch PPP-RTK user parameters. Supported by simulation-based results, our study demonstrates how a misspecified weight matrix can produce sub-optimal solutions, thereby corrupting both the ambiguity-resolved positioning solutions and the quality description that goes along with them.