We spend a lot of time with math leaders who oversee 20, 50, and even 200 schools. Nearly every conversation eventually arrives at the same question:
how do we get everyone in an entire system to strengthen their math instruction?
To be honest. It's the wrong question to start with.
At least, it's the question asked in the wrong order.
Before we can understand how to improve math in a system of 200 schools, we have to understand what it takes to change one math department, in one school, with one team of people who see each other every day. If we can't name the conditions that make change work at that scale, multiplying our effort by 200 doesn't get us change — it gets us 200 different versions of confusion, moving at 200 different speeds, for 200 different reasons.
Call it the math microcosm problem.
Get the small version right first, understand precisely why it worked, and only then ask how to hold that same shape at scale.
The six conditions every math improvement effort needs
When we break down what real, durable change requires — even in a single team — a few things show up every time.
A math coherence compass.
A four-point decision framework that includes a shared vision and a clear objective the group is genuinely working toward together, not just nodding along to in a meeting. Vision without a compass is a poster on the wall. A math coherence compass tells people which way is forward when they lose their bearings mid-week.
And before you file this under "nice idea, no time" — the thing we're describing has a research name and a measurable effect. Newmann, Smith, Allensworth, and Bryk (2001) called it instructional program coherence: a shared understanding of instructional priorities, materials, and pedagogical practices across a system. They found it was directly correlated with long-term student achievement gains. Coherence isn't the soft part of the work you get to after the real work. It is the work.
Clear, measurable outcomes.
Not aspirations. Not themes. Actual definitions of what "better" looks like, specific enough that two people looking at the same data would agree on whether it's working.
Commitment, not consensus.
In Farther, Faster, and Far Less Drama, Janice and Jason Fraser argue that teams get decisions wrong in two opposite ways: they move too slowly chasing full agreement, or they move fast but shallow, with no real buy-in — and both are forms of waste (Fraser & Fraser, 2023). Their alternative isn't consensus but consent. Instead of asking do we all agree? and is this the best possible option?, you ask two different questions: Does this move us forward? And can you live with it?
That's the difference between a decision that survives contact with a hard Tuesday and one that quietly unravels the first time someone objects.
In a school, that's less a philosophy than a habit: decide with your team, not for them. "We need to choose a content focus this year, let's decide together" builds something that "I've decided, help me plan the PD" never will.
Support structures that create epiphanies.
People don't change their minds because they were told to. They change because something challenged a belief they didn't know they were holding, and they had the time, safety, and opportunity to sit with that discomfort. If a system doesn't deliberately build in that time and opportunity and experience, the epiphanies simply don't happen — everyone stays exactly as convinced as they started.
Realistic milestones and key results.
Time-bound, measurable, sequenced so that each one builds momentum for the next. Big vision, small next win. Repeat.
Reduced fragility.
If the whole effort depends on one champion or two believers, it isn't a system change — it's a personality. The moment that person leaves, gets reassigned, or burns out, the work goes with them. Durable change means ownership and decision-making are distributed widely enough that no single departure can collapse it.
Why change is hard in even a single math department
Even holding all six of those conditions inside a single school, with a single department, is genuinely hard. Getting everyone in one room to actually share the same understanding of the goal — not a version of it, not their own interpretation of it, but the same goal — takes real, sustained effort. Eliminating ambiguity is exhausting work precisely because ambiguity is the default state of any group of people.
The time required is also larger than most plans account for. Research on professional learning points to roughly 49 hours of sustained, content-focused PD before instructional practice meaningfully shifts (Darling-Hammond et al., 2009). Most districts deliver less than half that — and of what they do deliver, only a fraction is spent on actual instructional improvement.
And then there's the relentlessness problem. Something else is always on fire. There's always a version of "we'll get serious about this next month, once things calm down." Things never calm down. The teams that actually change are the ones who refuse to let today's fire displace this quarter's work — not because they don't care about the fire, but because they've decided the fire doesn't get to set the pace for everything else.
What happens when you scale math improvement too early
Leaders often look at what it takes to shift one school and assume that scaling it is a matter of more communication, more training, more resources. But scale doesn't just add difficulty — it multiplies every point of fragility that already existed in the small version.
If your compass was fuzzy in one school, it's fuzzy in 200 different ways across 200 schools. If your milestones weren't specific enough to be unambiguous, you now have 200 separate interpretations of "on track." If your change depended on one or two committed champions, you don't have one point of fragility anymore — you have 200, and some of them will fail.
Michael Fullan and Joanne Quinn put it about as plainly as it can be put: "Fragmentation is the enemy of progress. A small number of core priorities, done well, will have more impact than a long list of disconnected initiatives" (Fullan & Quinn, 2016). Scaling a fragmented plan doesn't fix the fragmentation. It distributes it.
Richard Elmore spent decades studying instructional change and arrived somewhere similar from the other direction: real change happens "one teacher, one classroom at a time" (Elmore, 2004). You cannot mandate transformation across hundreds of classrooms simultaneously. Not because leaders lack authority, but because that isn't how belief moves.
You cannot bolt shared ownership, durable decision-making, and realistic milestones onto a system after the fact, at scale, all at once. Those conditions have to be understood, tested, and refined where you can actually see and feel them — in the microcosm — before you ask them to hold weight across an entire system.
The question math leaders should be asking instead
So the question isn't "how do we change 200 schools." It's: what are the actual conditions under which change works, for whom, and why — as observed in the smallest unit we can study closely? Once we can answer that with precision, scaling isn't about doing something fundamentally different. It's about protecting the same conditions, over and over, at a size where we can no longer watch every one of them happen ourselves.
Systems don't change because leaders declare that they will. They change when the conditions for change are true in enough microcosms, often enough, for long enough, that the system has no choice but to catch up.
Frequently Asked Questions
You're telling me the question is wrong, but I still have 62 schools and a board presentation in November. Are you asking me to shrink my scope?
No — you're being asked to sequence differently, not want less. The objective stays system-wide from day one. What isn't system-wide is the depth of support. Every school gets the direction and the decision-making framework. One gets the concentrated investment that produces visible proof.
For the board presentation, that's a stronger story than 62 schools of thin coverage: here's our one objective, here's the place we're proving it, here's what we'll be able to show you in June.
What actually counts as a "microcosm," and how do I choose which one?
A microcosm is the smallest unit where all the conditions for change can be seen operating at once — usually a grade-level team inside one building, or a small cluster of teachers sharing a coach. The test isn't size, it's observability. If you can't watch the conditions hold, it's too big to learn from.
Choose for readiness and visibility, not need. Look for a principal who will protect the time, a team with enough trust to be observed, and a location other people can actually visit. Pick the classrooms where your vision is already partly alive — not perfect, but real.
What about all the other schools while I'm doing this? What do I tell my superintendent?
They get clarity — the vision, the objective, and the same decision-making framework everyone else is using.
The honest principle underneath it: equity of attention is not the same as equity of outcome. Spread support evenly across a system that isn't ready for it and nobody gets enough to change. For your superintendent, the ask isn't patience — it's a different year-one metric. Not "how many schools did we serve," but "do we now have classrooms proving this is possible here, with our kids and our schedule?"
If I can't build all six conditions at once — and I can't — which one comes first?
The compass comes first, because it's the only condition you can hand to other people. Once principals, coaches, and PLC facilitators all evaluate decisions against the same four questions — are we moving toward our objective, does this reinforce our vision, does it align with our beliefs, can our system sustain it — leadership begins distributing itself.
One caution: shared language is not shared understanding. A district that rolls out a structure everyone nods at will still find teachers, coaches, and principals quietly working from three different pictures of it. Definitions, exemplars, and look-fors are what close that gap.
How long until I know whether it actually worked?
Expect two to three years for a team to reach proficient implementation, and roughly 49 hours of sustained, content-focused learning per teacher along the way. Teachers move through predictable stages — awareness, mechanical, routine, proficient — and most systems pull support at the mechanical stage, exactly when it's needed most.
You get signal much sooner. Choose one focus for six to eight weeks, align PLCs, coaching, PD, and leadership routines behind it, then look for movement rather than perfection. And apply the harder test: if you removed the coach tomorrow, would it hold? If not, you've proven a person works, not a system.
What happens to all of this when I leave?
If the work lived in your head, it stops. If leadership was genuinely distributed — shared decisions rather than delegated tasks, documented objectives and reflections, co-facilitated professional learning, relationships spread across the team — it continues through the transition.
One district in Eastern Ontario lost the math leader who had built their entire multi-year plan. By September the year's focus was set, schools were identified, coaching cycles were planned. It looked seamless. It was design. The second protection was that the team had publicly declared their two-year focus before she ever left, so her successor inherited a visible commitment instead of a blank page.
If any of this landed, our free Math Improvement Assessment is a good way to see which of these moves your own district most needs right now. It's on our site whenever you want a clearer read on where your year is really headed.
