The chemical abundance patterns observed in metal-poor Galactic halo stars contain the signature of the first supernovae, and thus allow us to probe the first stars that formed in the universe. We construct a theoretical model for the early chemical enrichment history of the Milky Way, aiming in particular at the contribution from pair-instability supernovae (PISNe). These are a natural consequence of current theoretical models for primordial star formation at the highest masses. However, no metal-poor star displaying the distinct PISN signature has yet been observed. We here argue that this apparent absence of any PISN signature is due to an observational selection effect. Whereas most surveys traditionally focus on the most metal-poor stars, we predict that early PISN enrichment tends to "overshoot," reaching enrichment levels of [Ca/H] ≃ − 2.5 that would be missed by current searches. We utilize existing observational data to place constraints on the primordial initial mass function (IMF). The number fraction of PISNe in the primordial stellar population is estimated to be <0.07, or ≲ 40% by mass, assuming that metal-free stars have masses in excess of 10 M☉. We further predict, based on theoretical estimates for the relative number of PISNe, that the expected fraction of second-generation stars below [Ca/H] = − 2 with a dominant (i.e., > 90%) contribution from PISNe is merely ~10−4 to 5 × 10−4. The corresponding fraction of stars formed from gas exclusively enriched by PISNe is a factor of ~4 smaller. With the advent of next-generation telescopes and new, deeper surveys, we should be able to test these predictions.