Every District Says Its Students Are Learning. This One Set Out to Prove It

The Return on Learning: Turning Assessment Data into Meaningful Student Outcomes

Assessment is moving from recall to demonstration. Mastery is becoming the unit of account. Portrait of a Graduate is redefining what a diploma certifies. West Hempstead stopped assuming its investments were working and built a district-wide system to show it.

‍

Background

Something has shifted in what districts are being asked to show. Saying students are learning no longer settles the question. Assessment is moving away from recall and toward demonstrated proficiency. Mastery of specific skills is becoming the unit of account rather than a course grade. Portrait of a Graduate frameworks are redefining what a diploma is supposed to certify. In New York, all of it is arriving at once. The state is rewriting its graduation requirements around project-based assessment tasks and demonstrations of proficiency, asking students to show what they know through writing, research, and complex problem-solving.

West Hempstead Union Free School District had already given that pressure a name. The superintendent talks about “Return on Learning” rather than return on investment—and he means it as a standard, not a slogan. Every initiative, every dollar, every hour of instructional time has to produce learning the district can actually demonstrate.

"’Return on Learning’ is really about being purposeful and intentional," explains Assistant Superintendent for Curriculum and Instruction Dina Reilly. "When we're investing in our children, we want to be able to demonstrate what they learned and how deep their knowledge goes."

"When we're investing in our children, we want to be able to demonstrate what they learned and how deep their knowledge goes." — Dina Reilly, Assistant Superintendent for Curriculum and Instruction

The standard was easy to state and hard to meet. The district had committed to one year's academic growth for every student. But when leaders looked at what they had to work with, they found a familiar problem. Khan Academy had been in the district for years, used informally by a handful of math and STEM teachers in whatever way each of them had worked out. No one could say which students were using it, in which subjects, or whether it was making a difference. It was a tool the district owned, not a practice the district ran. A resource that only some teachers use and no one can see into cannot produce evidence of anything.

So, when Director of Learning: ENL, Data & Student Success, Dr. Katie DiGregorio looked closely at Khanmigo, the question was not whether the tool was good. It was whether a district of roughly 1,660 students, with no dedicated technology team, could make it universal enough to generate real proof of learning. What followed was an 18-month build, from a small administrative pilot in spring 2025 to daily use by classroom teachers, ENL staff, academic intervention specialists, speech therapists, physical education teachers, and student teacher residents.

‍

Challenges

1. The bar moved from coverage to proof. Demonstrating that a student learned something, and how deeply, requires evidence at the skill level, collected continuously. Report cards and end-of-year exams do not produce it, and New York's shift toward proficiency-based graduation requirements raised the stakes on getting it.

2. A tool in the building, not in the instruction. Khan Academy had been used informally by some math and STEM teachers for years. That meant no coherence across content areas or grade levels and no visibility into whether any of it was working. The district could not measure a practice it could not see.

3. Proof requires everyone. One year's growth for every student cannot be demonstrated from a sample of curious volunteers. It meant every grade level and every content area, including students or teachers with no history of using the platform at all.

4. Adults were harder than students. Students adapted quickly. More seasoned teachers were skeptical not because they were indifferent but because new tools do not come naturally. A compliance mandate would have produced compliance, which is not the same as evidence of learning.

5. No dedicated team to carry it. With roughly 1,660 students and no AI or edtech department, implementation rested largely on one person. Reilly describes DiGregorio as the one doing "the heavy lifting," which meant the design of the rollout, rather than the size of the staff, had to do the work.

6. Any expansion had to be defensible. With national conversation intensifying around student screen time, adding technology to more classrooms required an instructional rationale the district could explain to families—and a way to show the time was buying learning.

7. The rollout surfaced a second problem. Once students had access, teachers discovered they did not know how to use AI. Students prompted the platform for answers rather than engaging with it as a learning tool, which meant the early usage data was not evidence of learning at all.

‍

Solutions

West Hempstead treated both halves of the problem as design work: build the adoption that generates evidence and build instruction that produces the kind of learning the new assessments will ask students to demonstrate.

• ‍ Set the standard out loud, and keep saying it. “Return on Learning” came from the superintendent and was repeated constantly, alongside the district's other refrain, “the relentless pursuit of excellence.” It gave every conversation about Khanmigo a shared test to meet: what did students learn and how do we know? Framing the work as an accountability standard rather than a technology initiative is what kept it from being received as one more thing with which they had to comply.

"It supplements strong instruction. It doesn't supplant it." — Dr. Katie DiGregorio, Director of Learning: ENL, Data & Student Success, West Hempstead UFSD

• ‍Train leadership before training teachers. DiGregorio spent the summer of 2025 on her own professional development, using Palm Beach County's Khan Academy training guides to build real expertise before bringing anything to staff. Administrators were then onboarded on opening day, focusing on the lesson-planning features, so they could support teachers from genuine understanding rather than encouragement alone. No teacher was asked to try something their principal could not discuss.
• ‍Phase professional learning by grade level  and ask for attempts rather than mastery. Grades 3-6 received mandatory half-day sessions organized by grade level, so teachers worked alongside colleagues who shared the same students, standards, and questions. Secondary teachers had sign-up PD periods and Model Schools days. The requirement was deliberately small: try Khanmigo a set number of times, in whatever way made sense for your instruction, with full support available. "Just try it!" became the refrain. This year, secondary school teachers will be trained in the features and uses of Khanmigo.
• ‍Let adoption spread sideways. A teacher who found a feature told the colleague next door. DiGregorio followed up with individual coaching. Administrators opened their calendars for drop-in sessions. Rather than a rollout radiating from a central office, a community of practice formed around shared experimentation.

"Once people sat down with Katie or when they turned to their colleague next door, it started to become something every grade level was talking about." — Dina Reilly

• ‍Design instruction for demonstration, not recall. As New York began publicizing its move toward project-based assessment tasks, teacher resistance dropped and design work started. A social studies teacher built a project-based unit for Global 9 using Khanmigo to scaffold research and writing. Writing Coach became significant at the secondary level. Elementary students used "Ignite My Curiosity" prompts to develop their own questions before engaging with content, building the metacognitive habits proficiency-based assessment demands. The district frames this as the path toward New York's Portrait of a Graduate competencies beginning in kindergarten, not in high school.

"They're beginning to leverage Khanmigo to redesign types of assessments in order to support students through that demonstration of proficiency—through writing, through research—in ways that meet each child where they are." — Dina Reilly

• ‍Evidence of mastery as a  base. Skill-level mastery data gave the district something a report card cannot: a continuous, granular record of what students can actually do. Teachers use features like "Practice my Knowledge" to get students to articulate their understanding of concepts rather than just producing answers. That sits alongside the district's hard data points, NWEA and RIT scores, Regents results, and state assessment data, with qualitative classroom observation on top. Khan Kids was deployed in the kindergarten building as a universal screener, making data-driven instruction possible from a child's first year of school. The district is also evaluating Khanmigo as a resource inside its MTSS/RTI process.
• ‍Teach AI literacy explicitly. When teachers noticed students trying to extract answers, they named it rather than restricting access. A building-wide assembly covered what AI is, what it is for, and what responsible use looks like. Beyond making the usage data meaningful, the work addressed New York State's K-12 digital literacy standards (digital footprints, responsible use, AI ethics) in an embedded way rather than as a separate checklist.

"We couldn't just assume they knew this. The teachers realized this was something they had to explicitly teach." — Dr. Katie DiGregorio, on teaching students responsible AI use

• ‍Keep technology in its place. Cell phones are banned during the school day. At the elementary level, students still use paper fraction strips, still write by hand, still present to an audience. Teachers are coached to ask what they are trying to do, what the intended outcome is, and why this tool, here, now. Khanmigo is used to enrich and assess that hands-on work, not replace it.
• ‍Extend past the classroom teacher. Academic intervention specialists, speech therapists, and physical education teachers integrated Khanmigo into their practice. For multilingual learners and students with IEPs, DiGregorio, who also directs ENL, describes it as creating entry points into lessons that did not previously exist. Student teacher residents from Adelphi University, who were placed in the district for a full year, use Khan Kids to drive action research and small-group instruction.

District-Reported Results

• ‍Evidence of learning. A resource used informally by a handful of teachers became a district-wide practice within 18 months, expanding from a spring 2025 administrative pilot to active use among PreK-12 classroom teachers, ENL staff, academic intervention specialists, speech therapists, and physical education teachers. Each of them now generates a record of what students are working on and where they are stuck. 
• ‍Measurable academic gains. NWEA data from fall to winter 2025 showed gains in lowest-risk student scores, attributed in part to targeted, data-driven use of Khanmigo to personalize instruction and monitor progress. 
• ‍Mastery became a shared language. Skill-level data gave teachers, specialists, and administrators a common way to talk about student progress that a letter grade does not provide. It also gave the district something concrete to put behind the phrase “Return on Learning.” 
• ‍Data-driven instruction from the first year of school. With Khan Kids serving as a progress-monitoring tool in the kindergarten building, the district began collecting actionable learning data before students are old enough to read a report card.
• ‍Teachers began asking for it. Word-of-mouth adoption led teachers to request coaching, explore new features, and train their own colleagues without being assigned to do so. In a rollout that started with a deliberately minimal requirement, the fact that teachers were exceeding the requirement  was the clearest signal that practice, not compliance, took hold.
• ‍Instruction is being redesigned for demonstrated proficiency. Teachers across content areas are building project-based learning experiences and writing-intensive tasks that map to New York's evolving graduation requirements and Portrait of a Graduate competencies and begin in kindergarten, well ahead of the state finalizing the framework.
• ‍AI literacy as a byproduct. A structured, building-wide approach to responsible AI use emerged from implementation rather than from a compliance requirement, authentically addressing New York State's K-12 digital literacy standards.
• ‍Expanding scope. The district is expanding its focus to include personal finance and entrepreneurship coursework and is using Khan Kids at the kindergarten level.

‍