SCIENCE OF TRUST
How We Measure Trust » The How
Each Lens Adds a Layer. One Prescription.
Structure, Alignment, and Trust, read together, produce Cohesion — a measurable answer to whether a community moves as one, or fragments. Here's what that actually looks like against a real question.
← Back to We Measure TrustThe Problem This Solves
The Conundrum
The same intervention. Two communities. Two different results — every time. The standard explanation is always the community: apathetic, resistant, hard to reach. The real variable has been invisible until now. It's trust — how much exists, where it's concentrated, and whether it transfers from one domain to the next. Once it's measured, the "conundrum" stops being a mystery and starts being a map.
Racine, Wisconsin reads as the most institutionally closed community in its district — that finding holds. What the citywide number hides is what matters more: trust doesn't move the same way in every neighborhood underneath it. Two miles from a neighborhood where trust barely converts to institutional engagement sits one that converts it three times better, on the same channels, in the same city.
Closed at the city level. Unevenly reachable underneath it. "Unreachable" describes a resolution failure — not a place.
The Method, Applied
Watch It Work
On the previous page we introduced the formula: Structure × Alignment × Trust = Cohesion. Here's what running that formula against an actual question looks like — each lens narrowing the picture until what's left is a decision, not a guess.
A constructed example built to show the method — not a documented case or client result.
The question: "A chronic-disease management program is launching across two neighborhoods in the same city. Which one will actually sustain participation past year one?"
What we find
Neighborhood A has three clinics and a hospital outreach office — dense, visible, institutional infrastructure. Neighborhood B has one understaffed clinic, but a dense layer of churches, mutual-aid groups, and barbershops that already move information block to block.
What this tells us
Neighborhood A looks like the easier launch — more formal infrastructure to plug into. Neighborhood B looks under-resourced by comparison. Structure alone would rank A ahead of B. It's the wrong ranking, and we don't know that yet.
What we find
Neighborhood A's clinics operate independently, each running its own outreach calendar with little coordination. Neighborhood B's informal network already coordinates naturally — the same families move between the church, the barbershop, and the mutual-aid group in a single week.
What this tells us
Neighborhood A's density isn't coordination — it's three separate efforts competing for the same attention. Neighborhood B's smaller network is already unified. The ranking from Structure alone starts to look shakier.
What we find
Neighborhood A shows high first-visit turnout at clinic events, but sharp drop-off after the first appointment — the behavioral signal of a community that responds to an invitation but doesn't yet trust the follow-through. Neighborhood B shows low response to anything posted at the clinic, but strong, repeat engagement with anything that moves through its existing relational network.
What this tells us
Neighborhood A's engagement is real but shallow. Neighborhood B's is scarce through institutional channels and strong through relational ones. The channel, not the community, was the variable all along.
What we find
Neighborhood B holds together as one network under a shared program, even though it looked the weaker candidate on paper. Neighborhood A's institutional density masks fragmentation — it will onboard well and lose participants steadily after.
What this tells us
This is the reversal the Conundrum predicts: the "harder" neighborhood is the one that sustains. Not because of any one person in it — because of how trust already moves through it.
The Prescription
Route Neighborhood B's launch through its existing relational network — churches, mutual-aid groups, informal community spaces — not through the clinic. Route Neighborhood A's launch through a single coordinated clinic calendar instead of three competing ones, and build a longer runway for follow-through. Neither fix required finding one specific person. What the read finds is the channel type that will actually carry the program — identifying the exact local leaders who can carry it is deliberate fieldwork that follows, done by people, in the neighborhood, informed by this read.
Takeaways
What This Shows
Principle 01
The obvious answer is usually wrong
Neighborhood A looked easier. It was actually the less sustainable choice. Neighborhood B looked under-resourced. It was actually the more cohesive network. You only know which is which by reading all three lenses together.
Principle 02
The channel is the method
Not all trust runs through the same kind of structure. The read tells you whether a place moves through institutions or through relationships — and routes accordingly. Finding the specific people who carry that relational trust is real fieldwork, done by humans, after the read — never a name the system hands you.
Principle 03
Launch is easy. Stewardship is the proof
The read finds where a program will onboard well and fade, versus where it will start slow and hold. That's the sustainability condition — found before a dollar is spent on outreach, not after year one's numbers come in.
On Privacy & Data Ethics
Everything above is derived from aggregate behavioral signals, read at census-tract precision. Zero personally identifiable information is collected, stored, or used at any stage. No individual is named, tracked, or profiled by the instrument.
The unit of analysis is always the community — never the person.
We believe this is the only ethical way to do this kind of work. It's also, as it turns out, a genuinely stable way to work: place-level architecture holds up without a customer file, and raises no privacy questions of its own.
One method, run against a different question every time — structure, alignment, and trust, read together, until what's left is a decision instead of a guess.
See the full toolkit →Each Lens Adds a Layer.
One Prescription.
Five variables. Five questions. Together they produce something no single question could — a complete picture of how trust actually moves in a community.
A constructed example built to show the method — not a documented case or client result. "Block A," "Block B," and "Miss Carmen" are composites, not a real place or a real person. We're using them to show how the five lenses build on each other — and, just as important, where the read stops and where real human fieldwork has to start.
Most tools ask communities what they think. CI asks five behavioral questions about how trust moves. Each question can only be asked after the prior one is answered. Skip one and the prescription is wrong.
What looks obvious at Phase 1 often reverses completely by Phase 2. The neighborhood that looks unreachable turns out to be the most elastic. The one that looks enthusiastic turns out to have no staying power. You only know which is which by running all five.
Block B: High inward orientation. They trust their neighbor. They don't trust the flyer on the telephone pole.
Answer
Launch through Block B, not Block A. Route the ask through a relational channel, not a flyer — in our story, that's Miss Carmen. Pair the school newsletter for Block A. Skip the channel with no current leverage. Build a stewardship role into the program design from day one. Without that, the garden lasts two years. With it, it's still there a decade later.
This is what five structural lenses produce that one survey question never could — and where the read hands off to real people doing real fieldwork.
Block A looked easy. It was actually the least sustainable. Block B looked closed. It was actually the most reachable. You only know which is which by running all five structural lenses.
Not all trust is portable, and the structural read tells us exactly which channels transfer and which don't. It never hands us a name — finding the actual Miss Carmen in a real neighborhood is deliberate, qualitative fieldwork that follows the read, not something the instrument outputs on its own.
Phase 4 finds the sustainability condition before a dollar is spent. A garden program without a built-in long-term stewardship role is not a program — it's a temporary installation.
Everything the instrument itself produces is derived from aggregate behavioral signals at the census-tract level. Zero personally identifiable information is collected, stored, or used at any stage. The instrument cannot name, identify, or count specific individuals — not Miss Carmen, not anyone.
The unit of analysis is always the community, never the person. Finding an actual person like Miss Carmen — in a real engagement, not this illustration — is separate, deliberate human fieldwork that follows a structural read. It's never a direct output of the data itself.
We don't own the community data we work with — nor do our partners. The community does. We believe this is the only ethical way to do this kind of work.
It's also, as it turns out, a genuinely stable and predictive way to work: place-level architecture holds up without a customer file and raises no privacy questions of its own.