Many autistic children require support to meet their learning needs. Given the heterogeneity within the autism spectrum it is plausible that different types of support might be better suited to different children. However, knowledge on what interventions work best for which children is limited. We examined the outcomes of autistic preschool-aged children receiving one of two community early intervention approaches. Our main objective was to understand which baseline child characteristics might be associated with the degree of individual response to intervention--whether prognostically (i.e., irrespective of intervention received) or predictively (i.e., specifically in the context of one or other EI approach). Participants comprised two matched groups of preschool-aged autistic children receiving either Group-based Early Start Denver Model (G-ESDM; n = 42) delivered in a 1:3-4 staff:child ratio or an Early Intensive Behavioural Intervention (EIBI; n = 40) delivered in combination of 1:1 and 1:2 staff:child ratio. Over an approximate one-year follow-up period, children in both groups made significant gains in Developmental Quotient (DQ) scores, and trend-level gains in adaptive behaviour composite scores. Higher attention to a playful adult measured via an eye-tracking task was prognostically indicative of better verbal DQ and adaptive behaviour outcomes for the cohort overall. Moderation analyses indicated a single predictive effect--of pre-program sustained attention for subsequent NVDQ outcomes specific to those children receiving G-ESDM. These findings suggest that fine-grained measures of learning skills offer promise towards the selection and tailoring of intervention approaches to meet individual children's learning needs.