You need several assessment cycles before you will be able to set percentages or set goals for change. If, say, you set as an assessment goal that 85% of majors will be accepted to PhD programs but the data shows after two years that only 25% of your students are meeting that goal, then 85% is likely untenable. Don’t set yourself up for failure. Perhaps 30% is a realistic goal that could be achieved in, say, five years. Set short-term and long-term goals: to reach 30% in five years you will plan for 1% growth annually. Remember that you are going to have to be able to justify how you came up with this benchmark number. There are only two ways to set benchmarks: external norms or internal data collection. You may not be able to set your benchmarks until you have had several cycles of data collection to identify a baseline.
Be open to the idea that the first tool you create to collect data might not work. Keep fine tuning it until you get the data you need. If you start asking different questions or decide a new skill is desired, don’t forget to go back and adjust the tool to collect data that will be useful.