The human capacity for technological innovation and creative problem-solving far surpasses that of any species but develops quite late. Prior work has typically presented children with problems requiring a single solution, a limited number of resources, and a limited amount of time. Such tasks do not allow children to utilize one of their strengths: their ability to engage in broad search and exploration. Thus, we hypothesized that a more open-ended innovation task might allow children to demonstrate greater innovative capacity by allowing them to discover and refine a solution over multiple attempts. Children were recruited from a museum and a children's science event in the United Kingdom. We presented 129 children (66 girls, M = 6.91, SD = 2.18) between 4 and 12 years old with a variety of materials and asked children to use those materials to create tools to remove rewards from a box within 10 min. We coded the variety of tools children created each time they attempted to remove the rewards. By comparing successive attempts, we were able to obtain insights about how children built successful tools. Consistent with prior research, we found that older children were more likely than younger children to create successful tools. However, controlling for age, children who engaged in more tinkering--who retained a greater proportion of objects from their failed tools in subsequent attempts and who added more novel objects to their tools following failure--were more likely to build successful tools than children who did not.