Researchers have typically described ABA session delivery by asking practitioners what they do. Here, we sought to contribute data to this conversation via 3.9 million ABA sessions spanning 5.42 years and 599 providers. We observed significant variability in how providers deliver services spanning skill acquisition trials per minute, correct and incorrect trials per minute, overall percentage correct within specific programs and across all programs, whether targets were custom or from a standard library, and the amount of behaviors targeted for reduction. We also analyzed how the above differed based on demographics, clinically-relevant variables, and potential staff contingencies. We failed to find a clinically-relevant reason why variability existed. However, the interaction between skill acquisition trial rate and the rate of behavior reduction targets resembled the dynamics of schedule-controlled behavior suggesting a functional relation. Future research will benefit from more complex modeling, analyzing varying levels of granularity, and incorporating temporal dynamics.