Nowadays, block-based programming environments are often used to offer a gentle introduction to learning a programming language. However, an assessment of students' programming skills based on the results of a programming task is not sufficient to determine all areas students are struggling with. We therefore introduce a learning analytics approach of measuring and evaluating the programming sequences of students that program with Scratch 3. With our measurement framework, it is possible to record, store and analyze programming sequences done on a publicly-available, instrumented Scratch 3 environment. Changes in the programming sequence are categorized regarding the used block types and types of program change. We conducted an exploratory programming trial with lower and upper secondary school students to investigate small-scale programming strategies in the recorded programming sequences. Our goals are to identify students in need of support and to identify recurring patterns used by students successful in the trial. Clustering with k-means makes it possible to identify struggling students based on both interacted block types and types of program changes. Recurring patterns in the programming sequences of successful students show that small-scale programming strategies are very diverse.