Comparing contrasting cases is a promising means to prepare learners for future learning from related direct instruction. The most prevalent type of preparation intervention used in this "case comparison approach" is providing contrasting cases together with comparison prompts. However, if the contrasting cases are complex learners might need more guidance than mere prompts in order to detect their relevant similarities and differences and thus exploit their full potential. The "inventing approach" entails having learners invent explanations that deal with the detected similarities and differences in addition to comparing contrasting cases. In this approach, however, it is unclear whether the generation of inventions per se has an added value to the concomitant detection of similarities and differences. Against this background, in a 2 × 2 factorial experiment with N = 80 eighth-grade students we varied the preparation intervention the learners received before they processed some instructional explanations according to: (a) "type of task while comparing contrasting cases" (responding to a comparison prompt vs. verifying provided comparisons), and (b) "inventing task after comparing contrasting cases" (with vs. without). We found that if there was no invention task, the learners who verified the provided comparisons learned more from the instructional explanations than those who received a comparison prompt. Requiring learners to generate inventions only after comparing the cases yielded differential results. The learners who detected only one similarity or difference among the cases benefitted from the inventing task, whereas the learners who detected all of the relevant features were hindered by it, albeit marginally.