Choosing school schedules means balancing many priorities.

School start and end times have ripple effects across a district, affecting students, families, teachers, and staff.


Here's how we helped SFUSD update school schedules.

Step 1: identify and model priorities and constraints

Re-aligning start and end times allows the district to fulfill several objectives, including:

  • Reducing transportation costs by consolidating times into tiers, allowing buses to be re-used more efficiently in the morning and afternoon.
  • Moving towards compliance with new California legislation (SB-328), which mandates later start times for teenagers.
  • Minimizing disruption with respect to schools’ existing schedules.
  • Ensuring that any change is equitable across the district.

Clearly, it is not possible to accomplish all of these objectives at once. For example, a high school starting at 7:45am must move to a later start time to follow new California standards, which entails some amount of disruption.

What optimization can provide is a way to understand the tradeoffs between different objectives as shown in the graph.






The district can then choose how to prioritize different objectives based on what is and isn’t feasible. Based on this analysis, SFUSD identified transportation savings as a top priority, but opted to reduce potential savings somewhat to minimize disruption to elementary schools. In addition, the district chose to align start times uniformly for teenagers, with all middle schools starting at 9:30 and all high schools at 8:40.

Step 2: find good solutions, collect feedback, and identify other concerns

Having determined the district’s priorities and their relative importance at a high level, the next step was to identify start times that followed these guidelines.

Importantly, at this stage, we generated not just one proposal, but thousands, each one aligned with the district’s high-level objectives. We then allowed district staff to interact with the proposals, and select the most appropriate one.

Why didn’t we just give one “best” solution? While optimization algorithms are a great way to explore a huge solution space, they are only as good as their input data, and do not have all the necessary context to make a decision. Providing many options with similar transportation savings and a similar amount of change for students and families is a way to put the final decision in the hands of policy makers who know what is best for the district, and not a complicated black box.