I was all nerves the night before my presentation at a big education leaders’ conference. But Iowa’s top education official didn’t even need to hear my pitch. I mentioned National Math Stars over drinks at the bar, and she immediately asked, “Can we get your program to Iowa this year?” Overnight, our small nonprofit had a chance to reach all of Iowa’s most mathematically talented children at once. We shook hands on it. If she helped with nominations, we’d support Iowa Stars.
That moment captured one reality of scaling: it’s not just about careful design, it also requires being ready when lightning strikes. Our pilot with 12 families taught us a lot about what worked and what didn’t, as I described in Part 1 of this series. But bigger challenges were ahead: growing thoughtfully, finding students efficiently, and seizing opportunities, all without losing focus or quality.
A state-by-state path to growth
As we set out to build our first full cohort of National Math Stars after the pilot, one of our initial strategic questions was geographic. Should we cast a wide national net, embed deeply in a single city, or take a middle path?
Casting a national net might sound ideal — more reach, more students — but programs that do this mostly find highly engaged, highly educated, middle-to-high-income families. And they’re concentrated in predictable areas, such as the San Francisco Bay, Manhattan, and Boston.
The other extreme is the “deep embed” model. Programs like Bridge to Enter Advanced Mathematics have done extraordinary work by building trust and infrastructure in specific neighborhoods. But for a program like ours, designed to serve students with “1 in 1,000” talent, no single city would have enough density.
So we chose a middle ground: a state-by-state approach. States are natural organizing units — they run education conferences, publish data, and often have infrastructure we could plug into as we worked to sign up individual school districts and schools. They also gave us enough geographic concentration to occasionally bring students together in person.
We aimed to find Stars who were mathematically extraordinary, unlikely to otherwise access similar opportunities, and diverse across many dimensions. When we scored states to see where these students were and where we could get good data to select them, Texas emerged as the clear winner. Our Head of Partnerships, Alex, ran with it.
In our first year (2024), we distributed 1,700 math awards to top-achieving Texan second- and third-graders, and identified 61 extraordinary students across the state to select as Stars.
Partner in finding students
The next question was how to efficiently reach mathematically extraordinary kids. We saw four options:
- Target parents directly
- Work with community-based organizations
- Partner with schools
- Secure state or national screening partnerships
The first three are “bottom-up” options that require an extensive ground game, and the fourth is a “top-down” approach that requires securing influential partnerships.
Most talent development programs try to reach parents directly. But in addition to skewing applications to well-resourced families, we’d get lots of untargeted applications since parents rarely know how their child compares mathematically to peers. We’d also miss many extraordinary students. Community-based organizations had similar limitations, and the landscape was too fragmented to serve as a consistent pipeline.
With top-down partnerships, the payoff can be enormous — it’s how the Brazil Math Olympiad has achieved stunningly high participation; and working with government has enabled some excellent experiments in talent identification and development in India. But if we built an identification strategy based on this approach and failed to get the right partners on board, our recruitment would collapse.
So we selected a set of states, and within those, we centered our outreach on school districts. Teachers, principals, and superintendents are trusted by families, and already have access to student performance data. More importantly, they could make targeted recommendations. School newsletters to parents go largely unread, but we learned with the pilot cohort that when a teacher or school leader tells a parent that an opportunity might be a good fit for their child in particular, it’s often taken seriously.
Seize surprise opportunities
Then came that chance conversation with the Iowa leader, State Department of Education Director McKenzie Snow. Until then, our identification strategy had relied almost entirely on school and district partnerships built one at a time. Suddenly, we saw what statewide nominating could do — and the results reshaped our view of what was possible.
Iowa was not initially part of our 2025 plan, but we jumped at the chance and quickly made it happen. Despite its small population, Iowa ended up with 17 Stars, the most of any of the six new states we added this year. This is a testament to the power of universal screening – and the potential for a visionary state leader to move quickly and get things done.
Onboarding a new state in a few weeks was a logistical challenge, but the results were stunning. We’re excited to partner with more state DoEs in 2026. But different states receive different math data at different times of the year. And state education leaders and their priorities change regularly. We can’t rely on top-down opportunities everywhere. So we’re enthusiastic about building both strategies in parallel: working with state DoEs while continuing our individual school and district outreach as well.
Seeking scale
Two years in, National Math Stars has grown from a pilot with 12 families to two programs serving over 400 Stars, selected from a pool of over 6,000 applicants. We’ve expanded into an independent nonprofit with 16 full-time staff and a $4 million annual budget. And that’s just the start.
Within a decade, we aim to distribute 80,000+ math awards each year (reaching at least half of the country’s top-2% of 2nd- and 3rd-graders) and to welcome cohorts of 4,000+ Stars annually.
We have plenty to figure out along the way. We’re currently piloting a new program model and wrestling with strategic questions: How quickly should we grow? Where do we go next? All this while rapidly scaling our team and infrastructure. The journey and learning that have come with it have been thrilling—and there’s much more ahead. We’ll be sure to keep GTF blog readers updated.
