Brianna Pollari is a Services Delivery Lead at Education Analytics, where she has led data partnership and analytics work with districts since 2019. Drew McDermott is a Senior Technical Program Manager of Data Science at Education Analytics, specializing in data infrastructure and on-track analytics since 2022.
At the start of 2026, Education Analytics (EA) closed out a five-year grant that became one of the most formative chapters in EA’s history. The work — supporting six Networks for School Improvement (NSIs) across California through the Bill & Melinda Gates Foundation — pioneered infrastructure that helped shape many of the products that EA builds today.
The Challenge: A Checklist Was Not Enough
Prior to the grant, EA used a checklist-based on-track rating system to help districts understand whether students were progressing toward graduation. The model was straightforward but rigid: a student either met a threshold or did not. There was no flexibility for a student’s strengths to compensate for areas where they still had room to grow.
We had the idea of a bespoke on-track points system for our California districts — something with much more flexibility. We wanted a weighting system where a student performing well in high-level courses — which matters enormously for graduation — could have that strength count, even if their attendance wasn’t as strong that quarter.
The result was EA’s On-Track point system: a 0-100 scale that placed students in one of five “lanes” based on a predictive model accounting for their credit accumulation, grades, GPA, attendance rate, strength of course-taking schedule, and, when available, test scores. The lanes ran from Highly Vulnerable — students at greatest risk of not graduating — to Postsecondary Competitive — students likely to graduate prepared for four-year college.
That model became the foundation for a five-year, Gates-funded initiative. EA’s role was to build, refine, and deliver the point system — not directly to districts, but through six intermediary organizations called Networks for School Improvement. The grant’s aim was explicit: to advance high school graduation and college success rates for Black and Latino students, and students experiencing poverty.
The Six Networks
EA supported six NSI grantees, all in California. The networks ranged from large research organizations to county offices of education to nonprofit change management consultancies. Each chose one or more of three focus areas: 8th grade on-track, 9th grade on-track, or college readiness (grades 11-12).
- BTSC (Breakthrough Success Community): A group of schools within CORE, California’s data consortium, focused on 9th grade on-track for college and career success with particular emphasis on math. EA had the longest-running relationship with BTSC, which became one of the most active NSI partners and helped anchor much of the early model development.
- CAED (California Education Partners): Focused on both 8th and 9th grade paths, working primarily with small and rural districts. CAED also supported early development of the 9th grade model.
- HTH CARE Network (High Tech High): A charter network focused on the 8th grade path, with a goal of increasing on-track rates for Black, Latino, Indigenous, and low-income students.
- AIR/LBUSD (American Institutes for Research, Long Beach): EA’s only single-district NSI, focused on 8th grade literacy within the Long Beach Unified School District.
- PSI (Partners in School Innovation): A nonprofit working with districts in the Eastside Alliance and LAUSD on 8th grade on-track outcomes.
TCOE CVNIC (Tulare County Office of Education): EA’s only college readiness NSI, focused on systems- and teacher-level improvement across schools in the Central Valley, with a goal of increasing the number of students who are college-ready on-track.
The Toolkit: Dashboards, Roadmaps, Analytics, and Iteration
To serve these networks, EA built a suite of tools anchored by a set of custom Looker dashboards — interactive, student-level visualizations that gave NSI partners and district/school leaders insight into where students stood. Of the 10+ dashboards created, two of the first remained the most popular: a network-level “Move Up & On-Track” view showing the distribution of students across the five lanes over time, and a “Student List” view that let educators drill into individual student data, including lane assignment, A-G points, GPA, attendance rate, and total points earned.
That was one of our first forays into standard BI dashboards that would serve many different districts with different needs. User research, row-level security, dashboards built on top of a data warehouse — the team was tackling all of it for the first time.
McDermott, who joined the project midway through, helped evolve the technical infrastructure. That included refining the on-track model — calibrating it in close collaboration with EA’s research team, then working with the Data Engineering team to translate updated model parameters into student-level calculations flowing through the data system, which were automatically calculated as new data was shared.
Also utilizing the on-track model were EA’s College Roadmaps, created in partnership with CORE. For participating NSI districts, EA delivered student-facing roadmaps that broke down students’ college readiness into tangible pieces, including their on-track lane, a course-taking tracker, and personalized action items.
EA also managed the grant’s reporting requirements on behalf of all participating districts, calculating and compiling Gates-required metrics for each NSI and submitting them to Double Line, the company contracted by Gates to analyze results across networks. EA also submitted data to the National Student Clearinghouse for postsecondary enrollment tracking and provided reporting to Mathematica for the cross-California NSI evaluation. EA’s data expertise and capacity saved significant time for the NSIs and districts in managing the data collection and reporting themselves.
The NSIs needed more custom metrics and the Looker dashboards were not fully meeting all districts’ needs, so EA pivoted. Around year three, the team introduced tracking and “common analytics” — a menu of standardized reports and data extracts districts could request each year. Offerings included attendance extracts, course extracts, math-specific GPA breakdowns, and Sankey plots showing year-over-year grade-level course changes.
“The main focus of the grant was to support these NSI districts,” Pollari said.
The Ed-Fi Chapter: Lessons That Lasted
The NSI project also included one of EA’s earliest attempts at Ed-Fi implementation — the open-source data standard that now sits at the core of EA’s product suite. The lessons it produced were invaluable.
The contrast with later implementations was stark. In South Carolina, where the state mandated Ed-Fi onboarding and led the effort directly, the process was far smoother. In Texas, EA’s strong relationships with regional service centers made a similar difference. For the NSI project, the intermediary structure created a layer of distance — and the districts, unfamiliar with EA, did not always understand why the work mattered.
There was also a proportionality problem. “We were asking districts to do hours of mapping to support only the select number of their schools participating in a temporary grant.” EA has since refined its onboarding approach to ensure the districts doing the work are also the ones who benefit directly from it.
What made this project successful was that we were always working toward something concrete — a student’s chances of graduating, of getting to college. The data wasn’t abstract. It was about whether a specific student in a specific school was on track, and whether the adults around them had what they needed to help. That’s why we do this work.
The Through-Line: From NSI to EA’s Product Stack
EA’s current product suite still carries the NSI project’s blueprint.
The Looker dashboards were a direct precursor to Podium. The user research practices, rigorous row-level security and data access practices, the challenge of building standardized dashboards for districts with different needs, the balance of showing district, school, and student-level information — all of it was worked on, for the first time, in the NSI context, and something EA continues to develop and refine.
The data infrastructure underlying those dashboards was “Version 1.0” of what became Stadium. Modeling Ed-Fi data into standardized, analytics-ready tables in Snowflake — building a BI data model from structured data flowing out of a common standard — was tried out for the NSI work before it became the architecture for EA’s flagship platform.
What the Data Showed — and What It Meant During COVID-19
EA’s work with the NSI networks also produced important early insights about student outcomes during the pandemic. In 2021, EA published an analysis across more than 22 California districts, finding that the share of students in the “Highly Vulnerable” lane had approximately doubled — from around 15 percent in a typical year to roughly 30 percent in 2020-21.
The culprit was not course rigor. Students were taking similarly demanding schedules. What had shifted were grades and attendance: the proportion of students with GPAs between 0 and 1 nearly doubled in some grades, and chronic absenteeism spiked. A significant group of students who would have been in the middle of the on-track distribution in prior years had slipped into the most vulnerable category — a finding that helped districts understand what they were facing and where to direct their recovery efforts.
What Success Looked Like
Success on this project wore a lot of faces. Sometimes it was a district coordinator pulling up a Looker dashboard the night before a staff meeting, ready to make the case for a student who needed intervention. Other times it was an NSI partner emailing to ask for a change to an extract — they were about to use it in a meeting and needed it to reflect something specific.
Success also looked like CORE, after EA wound down its direct support, choosing to add its own on-track metric to their dashboard using the same point system EA had developed — sustaining the infrastructure independently because the work had been valuable enough to continue.
The Evaluation Is In — and the Results Are Promising
EA closed out this work just as Gates and their research partners — RAND, AIR, and Mathematica — released their final evaluation reports on the broader NSI initiative. Both McDermott and Pollari are credited in the Mathematica final report for cleaning and analyzing the California administrative data that underpinned the evaluation.
The findings were substantive. The 9th grade on-track NSIs produced consistent, positive impacts on course-related outcomes across four years of participation — improving student GPA by roughly 0.11 points on a 4-point scale and course pass rates by 3 to 4 percentage points. The 8th grade NSIs improved math test scores by the equivalent of moving a student from the 37th to the 40th percentile. And the postsecondary enrollment NSIs increased FAFSA completion rates by 3 to 5 percentage points — meaningful gains, particularly given the disruptions of COVID-19.
The evaluation also found that the approach worked best when schools engaged deeply with both their network and with continuous improvement practices — and when they centered equity explicitly in their work. That matches what the EA team saw on the ground: the relationships with districts mattered as much as the dashboards.
The full Gates blog post and links to all evaluation reports are available at the Gates Foundation’s website.
