The purpose of this study is two-fold: 1) to investigate the effects of Data-Driven Learning (DDL) framed within the Involvement Load Hypothesis (ILH) on Thai learners' use of academic collocations and 2) to examine how Thai learners utilized the involvement load (IL) components (need, search, and evaluation) to master academic collocations. It is hypothesized that the more learners actively engage with the IL components in a DDL task, the greater their development will be in acquiring new collocations. Participants (n = 31) were provided with repeated exposure to authentic L2 input (COCA) through DDL. They explored and analyzed three collocation patterns: adjective-noun, verb-noun, and adverb-past participle. The learners also verbalized their thoughts while completing the DDL tasks. A multiple-choice test and a collocation judgement test were constructed to assess the students' academic collocation competence. The results demonstrated that the ILH-based DDL generated a significant effect on Thai learners' performance of academic collocations. Thai learners also utilized all of the IL components in acquiring academic collocations through DDL. The search component was used most frequently, followed by the evaluation and need components. These findings provide useful insights into both the predictions of the ILH and L2 pedagogy regarding academic collocation learning.