What CI DOES — And Doesn't (Yet)

We are a lab. We are always learning and evolving decision tools and services from rich work in practices areas. We work with partner to solve challenges in their communities. The more we learn on each engagement the more sharper our tools and services become.

This piece outlines where we are heading.

What Community Intelligence Actually Does · RFUTR
RFUTR Inc. · 501(c)(3) · Washington, DC

What Community Intelligence Actually Does — And Doesn't Yet

A grantmaker-facing overview of a behavioral trust-mapping instrument, built to be read alongside your own diligence process, not around it.
The Problem We Built This For

Every funder placing dollars in a community is making a bet about whether that community will actually use what gets built. Almost none of that bet is informed by data — because the data that would inform it doesn't exist in the standard toolkit.

Demographic data tells you who lives somewhere. Polling tells you what people say they'll do. Neither tells you how trust actually moves through a place — who a community listens to, which institutions it has already written off, and where the social fabric is strong enough to carry a program past its launch. Community Intelligence (CI) was built to close that specific, narrow gap: not to replace program judgment, but to give it a ground-truth layer it currently lacks.

What It Is

Census-Tract Precision, Not a Survey

CI reads structural trust patterns at census-tract resolution — roughly 4,000 people per tract — using behavioral indicators drawn from public and administrative records, not surveys and not individual profiling. It collects zero personally identifiable information at any stage. It cannot name, identify, or count specific people. What it produces is a structural read of a place; identifying the actual people who carry trust in that place is human fieldwork the read informs, not something the instrument does itself.

The current architecture tracks roughly 350 indicators, of which 140–145 are typically populated for a given census tract, across seven civic domains carrying tract-level signal (two of nine originally scoped domains — Governance and Safety — are anchored to city-level values only and do not yet have tract resolution; we say so rather than blur the two). Inputs refresh on an annual cycle to five-year rolling averages — this is not a live sensor network, and we don't describe it as one.

Where CI sits competitively: it's a third category, distinct from polling and prediction markets. It asks no one anything, and no one profits from the answer. Individual-level data can outperform CI where it's available and consented to — we don't compete there. Place-level architecture is the part that's stable without a customer file and raises no privacy questions on its own.
Capability Honesty

Tested & In Active Use — vs. — In the Lab

Every instrument we name carries a stated test status. Where nothing has been tested, we say "tested against: none" rather than imply otherwise. That standard governs everything below.

Tested & In Active Use

  • Floor Identification — reliably finds the most institutionally disconnected places in a district. Validated against real turnout outcomes (see Proof, below).
  • Trust Flow mapping & Dark Spots — structural reachability variance within a single geography, independent of poverty.
  • Institutional-gap divergence — comparing what institutions market against what neighborhoods actually measure.

In the Lab

  • Gradient ranking — ordering places by degree of connection, not just floor vs. not-floor. Tested and did not hold; not offered as a capability.
  • Trust Elasticity / Trust Resilience — tested at essentially no measured relationship to behavior; not claimed as predictive.
  • Sustainability / stewardship prediction — CI can find which channel is likely to carry a program; whether it sustains is a tracking question during the engagement, not a pre-spend prediction.

A null or failed test is information we report, not a result we bury. That is a deliberate methodology stance, not a gap we're hoping you won't ask about.

Proof, Not Promise

Two Anchors, With Real Numbers

The Election Day Test (Floor Identification)
A sealed, dated, countersigned prediction — made before any outcome data existed — that CI could identify the most institutionally disconnected neighborhoods in a district. 3 of 4 pre-registered predictions held against actual turnout; 1 did not, and we report it. The instrument reliably finds the floor. It does not reliably rank places by degree of connection — that capability was tested separately and failed, and we don't claim it.
Houston — the Institutional-Gap Divergence
Institutions in Houston market Achievement, Self-Direction, and Stimulation as what the neighborhood values. The neighborhood-level read measures something different: 90.5% Loyalty, 85.3% Care, and 53.5% functioning in a guardian role — against an institutionally claimed 25%. A 28.5-point gap between what gets marketed and what actually holds a community together. This is our strongest current verified finding and it hasn't needed softening.
Where This Fits a Grantmaking Process

Three Uses, Not One Product

UseWhat It Adds
Pre-grant due diligenceA structural read of whether a proposed geography has the trust infrastructure to carry a place-based grant — before commitment, not after.
Cohort / portfolio selectionComparing candidate communities on reachability and institutional-gap variance, with a stated coverage rule for which geographies are actually classified.
Community-first digital & civic tech evaluationReading whether a proposed digital tool is being built from an existing trust network in a place, or imposed on one — directly relevant to funders wary of novelty-first tech bets.
What We Will Not Tell You

The Honest Boundary

CI will not tell you whether a specific program will succeed — it tells you about the trust terrain the program will land on. It will not name or count individuals. It will not rank places by degree of connection with confidence we don't have. It will not produce a "89% accurate" headline number — no such study exists, and we won't manufacture one to make a pitch land. Where a geography isn't yet classified, we say so plainly rather than imply national coverage that doesn't exist yet.

This is not a deficit framing. It's the same discipline we'd want from any instrument asking a funder to change how it places capital.

RFUTR Inc. is the 501(c)(3) that carries this work into civic and place-based practice — non-electoral, structurally separate from our commercial licensing partners. We'd rather show you where the instrument is strong, where it's still R&D, and let that honesty be the pitch.

Previous
Previous

COMMUNITY TRUST MEASURE

Next
Next

TRUST window