Bull Moose Strategy LLC — Launch Whitepaper Summer 2026 Beta Release
Executive Summary
Republican down-ballot candidates — county commissioners, township trustees, school board members, tax levy opponents — have been priced out of the voter data tools that win races. NationBuilder costs $150-500 per month and is built for statewide campaigns. GOP Data Center requires party apparatus approval. L2 Data sells commercial bundles priced for legislative and congressional clients. The county commissioner race with a $15,000 total budget has nowhere to go.
BMS Voter Intel closes that gap.
Built on 18 public data sources — federal donor records, parcel data, farm subsidy filings, census enrichment, precinct election results, and more — BMS Voter Intel delivers custom voter universes, automated digital ad audiences, and field-ready contact lists at a price point matched to the race, not the party's preferred vendor.
What it does: Ingests, cross-references, and scores a county voter file against 18 enrichment layers. Produces strategic universes (Hard-R GOTV, Soft-R Turnout, Unaffiliated R-Leaning Persuasion, Cross-Party Swing) and pushes hashed-PII audiences directly to Meta Custom Audiences and Google Customer Match.
Who it is for: Republican and independent candidates at county level and below — commissioner, trustee, school board, municipal, judicial. Races where the candidate is also the treasurer and the field director.
Why it is different: No party gatekeeping. No bloated SaaS subscription. No vendor relationship required. Every data source is public record. Every universe decision is transparent. The platform runs on $6.50 per month in infrastructure.
When it launches: Limited beta, Summer 2026. Capacity is intentionally constrained to maintain data quality and client service standards.
What it costs: Contact for pricing. No tiered software subscriptions. Engagement-based model matched to campaign scope.
The Problem
NationBuilder: Built for Organizations That Are Not Your Client
NationBuilder's entry price is $29 per month for 2,000 contacts. That sounds accessible until you realize a single county voter file — Licking County, Ohio, for example — contains 112,000 registered voters. At that scale, NationBuilder pricing runs $149 to $499 per month. Annually, that is $1,800 to $6,000 for the data layer alone — before staff, before ads, before mailers.
For a county commissioner race with a $15,000 total campaign budget, that is 12 to 40 percent of the entire campaign spent on a platform subscription.
NationBuilder was designed for nonprofit advocacy organizations, statewide candidates, and parliamentary parties in Westminster democracies. Its feature set reflects that: national party integration, complex tagging taxonomies, blog hosting, donor management, email broadcast. The township trustee running in a 4,000-voter precinct does not need any of that. She needs a GOTV list segmented by precinct and a Facebook audience targeting soft Republicans in her specific township.
NationBuilder cannot sell her that at a price she can pay.
GOP Data Center: The Party's Database, on the Party's Terms
GOP Data Center (formerly the Republican National Committee's VAN equivalent) provides voter data at competitive price points — but access flows through state and county party committees. If the candidate has a relationship with the party establishment, this works. If the candidate is running in a contested primary, challenging an incumbent, or operating in a county where the party committee is indifferent or hostile, access is uncertain.
Data access should not be contingent on party committee goodwill. Down-ballot candidates deserve the same data infrastructure as the party's preferred pick.
L2 Data: Commercial-Grade Pricing for Commercial-Grade Clients
L2 Political is the gold standard for modeled voter data: consumer data overlays, vehicle registration, magazine subscription modeling, and sophisticated propensity scores. It is also priced for legislative caucuses, statewide campaigns, and political consulting firms operating at scale. The per-record and per-pull pricing structures are not designed for a single county race.
L2 is an excellent product for the clients it was built for. County commissioner races are not those clients.
The Vacuum
Below the state legislative level — county commissioner, township trustee, school board, municipal judge, fire levy campaign — there is no well-resourced, accessible, affordable voter intelligence product designed specifically for the race type. Candidates either pay NationBuilder rates they cannot afford, depend on party committee goodwill they may not have, or run blind on the county's raw voter file with no enrichment, no scoring, and no integration with the digital platforms where the campaign actually runs.
BMS Voter Intel was built to occupy that vacuum.
What BMS Voter Intel Does
Universe Construction
Every voter file starts as a list of names, addresses, and party registrations. BMS Voter Intel transforms that list into a scored, segmented, and enriched intelligence product.
The platform ingests a county voter file and cross-references it against 18 public data sources. The result is a voter record that includes: property ownership status, farm program participation, small business ownership, federal and state political donation history, union leadership role, probabilistic ethnicity, inferred age cohort, socioeconomic neighborhood profile, proximity to partisan-correlated retail locations, precinct-level swing history, and a multi-factor swing voter score.
From that enriched base, the platform builds named strategic universes:
- Hard-R GOTV — reliable Republican voters who need turnout contact only. High-value, low-persuasion-cost.
- Soft-R Turnout Push — registered Republicans with inconsistent turnout history. Responds to motivation messaging.
- Unaffiliated R-Leaning Persuasion — unaffiliated voters who show Republican-correlated signals across property, donor, and behavioral layers. Primary target for persuasion.
- Cross-Party Swing — voters with cross-party signals: donation to opposite party, cross-party primary history, precinct flip history. Requires careful evaluation before inclusion.
Each universe is built for a specific campaign use case. A GOTV campaign in the final 72 hours uses different lists than a persuasion program running six weeks out.
Digital Audience Push
Universes are not just for field programs. The same voter records power digital advertising.
BMS Voter Intel generates hashed-PII customer lists from each universe — name, email, phone, address run through the same SHA-256 hashing required by Meta and Google — and pushes them directly to:
- Meta Custom Audiences (Facebook and Instagram): Match rates of 29 to 40 percent observed on Licking County voter file
- Google Customer Match: Match against signed-in Google accounts for Gmail, YouTube, and Search targeting
The result is a digital audience built from voter data, not from demographic proxies. A campaign targeting Unaffiliated R-Leaning Persuasion voters does not rely on Facebook's interest targeting approximations. It targets the specific voter records the universe identified.
Field List Delivery
Universes are delivered as field-ready exports: name, address, precinct, phone (where available), and all scoring signals. Walk lists can be generated by precinct, sorted by street. Phone bank lists are deduplicated and prioritized by score tier.
FCC Political File Monitoring
The platform ingests FCC Political Files — the broadcast ad-buy filings that federal law requires all stations to publish. This functions as an early-warning system: when an opponent begins buying broadcast airtime, the filing appears in the FCC database before the first spot airs. Campaigns using BMS Voter Intel get advance notice of broadcast activity, buy totals, and station selection — without waiting for the opponent to release numbers.
The 18-Source Data Stack
BMS Voter Intel is built on 18 public data sources, all of which are legally accessible public record. No consumer data is purchased. No social media scraping is used. Every source listed below is a government record, a federally mandated disclosure, or a publicly funded research database.
The following table describes each source, what signal it adds, and why it matters to a down-ballot campaign.
Federal Political Donor Records
Source: FEC itemized contribution disclosures (OpenFEC API)
Federal election law requires campaigns to disclose all contributions of $200 or more. The FEC publishes these records — contributor name, city, zip, employer, occupation, amount, recipient — going back to the 1980s.
For Licking County, Ohio: 2,363 registered voters matched to FEC contribution records. Each match is tagged with recipient party, contribution amount, and election cycle.
Signal: A voter who has written a check to a federal Republican candidate is the most reliable Republican signal in the dataset. A voter who has donated to both parties in different cycles is a strong swing indicator.
Ohio State Campaign Finance Records
Source: Ohio Secretary of State campaign finance disclosures
Ohio law requires disclosure of state and local political contributions. The SOS publishes searchable contribution records covering state legislative, statewide, and county-level races.
For Licking County: 766 registered voters matched to state campaign finance records.
Signal: State-level donation behavior often captures local political identity that federal FEC records miss — a township trustee donor who has never given federally, a school board opponent who funded a judicial campaign.
County Parcel Records
Source: Licking County Auditor parcel database
County auditors maintain property ownership records as part of tax administration. These records include owner name, mailing address, property address, acreage, land use classification, and parcel geometry (GIS boundaries).
For Licking County: 80,848 voter-parcel matches identified (owner-occupant flag applied to filter absentee landlords).
Signal: Owner-occupancy is a significant correlate of community investment and conservative economic interest. Rural acreage and agricultural land classification add farm-economy signal. Parcel geometry enables PostGIS spatial joins — walking distances, precinct overlays, township boundaries.
SBA Paycheck Protection Program (PPP) Loan Records
Source: SBA FOIA-released PPP loan database
The SBA released the full PPP loan recipient database under FOIA. The release includes borrower name, city, zip, loan amount, business type, and lender.
For Licking County: 578 registered voters matched as sole-proprietor or small-business PPP borrowers. Observed partisan lift: 2.5x Republican over county baseline.
Signal: Small business ownership — even informal sole-proprietorship — is a strong predictor of Republican alignment on economic issues. This is the most reliable small-business-owner signal available in public record.
USDA Farm Subsidy Records
Source: USDA Farm Service Agency payment records (EWG Farm Subsidy Database)
USDA discloses farm program payments under the Freedom of Information Act. The Environmental Working Group maintains a searchable database of recipients by name, county, and program.
For Licking County: 604 registered voters matched to USDA farm program recipients. Observed partisan lift: 2.4x Republican over county baseline.
Signal: Farm program participation is a near-perfect agricultural economy indicator. In Ohio's rural and exurban counties, farm-family voters are among the most reliable Republican constituencies. Identifying them precisely enables highly targeted agricultural economy messaging.
DOL LM-2 Union Officer Filings
Source: Department of Labor OLMS union financial disclosure reports
Federal law requires labor unions to file annual financial disclosures (LM-2 reports) listing named officers and their compensation. These records are public.
For Licking County: 777 registered voters matched to union officer roles. Observed partisan lean: 1.9x Democratic over county baseline.
Signal: Named union officers are a reliable Democratic-lean indicator. This is used to remove high-probability Democratic voters from persuasion universes and flag them for opponent-support risk scoring.
Census 2010 Surname Database
Source: U.S. Census Bureau Frequently Occurring Surnames database
The Census Bureau publishes a database of surnames appearing 100 or more times in Census 2010, with associated race/ethnicity probability distributions derived from census responses.
Coverage: 100% of voter records (probabilistic assignment at surname level).
Signal: Probabilistic ethnicity estimation at the surname level. Used for demographic composition analysis of universes and for compliance monitoring of audience diversity distribution. Not used as a hard filter — treated as a probability weight.
SSA Popular Baby Names
Source: Social Security Administration annual baby name frequency tables
The SSA publishes annual tables of the most popular baby names by year and sex from 1880 forward.
Coverage: 100% of voter records (probabilistic age inference from first name + registration birth year).
Signal: First-name era inference. Used to cross-validate registered birth years and to identify likely age cohort when birth year is absent or suspect. Secondary signal for generational messaging segmentation.
Partisan-Correlated Retail POI Distance
Source: OpenStreetMap Points of Interest
OpenStreetMap provides freely available point-of-interest data including retail locations by name and category.
Applied chains: Cracker Barrel, Hobby Lobby, Tractor Supply (Republican-correlated); Whole Foods, Trader Joe's (Democratic-correlated). Drive-time and straight-line distance calculated for each voter record using PostGIS.
Signal: Retail proximity is a documented correlate of partisan lean at the precinct level. A voter living 2 miles from a Cracker Barrel and 45 miles from a Whole Foods carries a different baseline prior than the reverse. Used as a neighborhood-level prior, not a deterministic signal.
Census ACS 5-Year Estimates (Block Group)
Source: American Community Survey, 5-year estimates, block group geography
The ACS produces the most detailed small-area socioeconomic data available in the United States. Block groups average approximately 1,500 people — sub-precinct resolution in most Ohio counties.
19 variables applied per voter: median household income, median home value, owner-occupancy rate, educational attainment tiers, employment sector mix, commute mode, housing age, household size, and others.
Signal: Neighborhood socioeconomic profile. Used to distinguish within-precinct variation where individual-level data is sparse and to validate or contradict individual-level signals.
ZCTA Boundaries + IRS ZIP-Level Income Statistics
Source: Census ZCTA shapefiles + IRS Statistics of Income (SOI) ZIP-level data
The IRS publishes annual ZIP-code-level statistics derived from actual tax return filings — not survey samples. These include AGI distribution, number of filers, deduction types, and business income prevalence.
Signal: ZIP-level income distribution from actual filer data. More reliable than ACS income estimates for affluent ZIP codes where survey undercoverage is documented. Provides full-filer income brackets per voter ZIP.
Household Co-Residence Inference
Source: Voter file address matching (internal)
Voters sharing a mailing address are inferred as household co-residents. Party registration and scoring signals are propagated within households using a weighted averaging model.
Signal: The single most reliable individual-level predictor for an unaffiliated voter is the party registration of the person at their address. A spouse registered Republican is a stronger prior than most demographic signals. Co-residence inference captures this without requiring purchased consumer data.
Precinct Election Results 2016-2024
Source: Ohio Secretary of State certified election results; Licking County Board of Elections
Per-precinct results for presidential, gubernatorial, senatorial, and county commissioner races from 2016 through 2024 — four full election cycles.
Signal: Precinct partisan margin, trend direction, and flip history. A precinct that voted +12R in 2016 and +4R in 2024 is moving. A precinct that flipped from R to D in 2020 and back to R in 2022 is structurally competitive. This history is the foundation of geographic targeting decisions.
5-Factor Swing Voter Indicator
Source: Derived — voter file + donor records + precinct results (internal composite)
The swing voter indicator is a composite score built from five observable behaviors: cross-party primary participation, turnout inconsistency across cycles, cross-party donation history, recent voter registration, and residence in a precinct with flip history.
Signal: The swing score identifies voters most likely to be genuinely persuadable — as distinct from low-propensity partisans who simply vote inconsistently. The construction logic and weighting are proprietary. The output (scored 0-100) is delivered to clients.
IRS County-to-County Migration Data
Source: IRS Statistics of Income county-to-county migration flows
The IRS tracks address changes reported on consecutive tax returns and publishes county-to-county migration flow tables annually. Each flow record includes number of returns (households), number of exemptions (individuals), and aggregate AGI.
Signal: Recent in-migration from high-Republican-index counties strengthens R-prior for new registrants. Recent in-migration from high-Democratic-index counties (e.g., Franklin County, Cuyahoga County) strengthens caution for unaffiliated registrants from those origin counties. Used to adjust scoring priors for voters registered within the past four years.
Census State-to-State Migration
Source: Census Bureau state-to-state migration flows (ACS migration tables)
The Census Bureau publishes annual state-pair migration tables derived from ACS. These capture gross and net migration flows between all state pairs.
Signal: State-of-origin context for voters who relocated from out of state. Used to calibrate county-to-county IRS signals and to contextualize new registrant behavior in fast-growing exurban counties.
UW Wisconsin Net Migration Cohorts
Source: Applied Population Laboratory, University of Wisconsin — 70-year age-cohort migration database
UW's Applied Population Lab maintains historical migration cohort data covering 70-year age bands, allowing reconstruction of when different age cohorts arrived in a county.
Signal: Long-term population stability context. Used to distinguish established multi-generational county residents from more recent arrivals — a relevant distinction in counties experiencing suburban-to-rural migration from urban cores.
FCC Political Files
Source: FCC Political File public database (required disclosure under 47 CFR § 73.3526)
Federal law requires broadcast stations to publish all political advertising purchase contracts within one business day of execution. The FCC's online database is publicly searchable by market, station, and candidate/committee name.
Signal: Real-time opponent broadcast activity monitoring. When an opponent buys broadcast time, the FCC filing appears before the first spot airs. BMS Voter Intel monitors relevant DMA markets for candidate and committee name filings. Early-warning capability for campaign resource allocation.
How It Works
Architecture Overview
BMS Voter Intel runs on a hybrid architecture: Neon Postgres (PostGIS-enabled) for live querying and universe construction; Cloudflare R2 for voter file archiving and large export staging; Cloudflare Workers for the API layer and automated audience push orchestration.
Total infrastructure cost at current Ohio deployment scale: approximately $6.50 per month.
Multi-Tenant Row-Level Security
Each client engagement is scoped to a named universe. PostgreSQL Row-Level Security (RLS) policies enforce universe boundaries at the database level — not in application code. A client querying the Licking County voter file through their campaign portal cannot access records outside their assigned universe, regardless of query construction.
This architecture supports multiple simultaneous client engagements on the same underlying county voter file without cross-client data exposure. It is the same multi-tenant isolation model used in enterprise SaaS applications.
Hashed-PII Compliance for Digital Audiences
Meta Custom Audiences and Google Customer Match both require that voter PII be hashed before upload. The platform applies SHA-256 hashing to name, email, phone, and address fields in accordance with Meta's and Google's Customer Match specifications before any data leaves the BMS infrastructure.
No raw voter PII is transmitted to Meta or Google. The match process occurs on Meta's and Google's side using their own hashed first-party data. BMS never receives match confirmation at the individual voter level — only aggregate match statistics.
Observed match rates on Licking County voter file: 29 to 40 percent across multiple audience pushes (Meta, varying by universe composition and data recency).
API Access and Client Portals
The platform exposes a Cloudflare Worker API with JWT-enforced access control. Client campaign portals can query universe composition, pull field list exports, and trigger audience push jobs through authenticated API calls. All access is logged at the request level.
Use Cases
1. R Primary GOTV — Final 72 Hours
Scenario: A county commissioner candidate in a four-way R primary needs to maximize turnout from reliable Republican voters in the final three days.
Platform role: Pull the Hard-R GOTV universe for the candidate's geographic scope. Generate precinct-sorted walk lists for volunteer deployment. Push the same universe as a Meta Custom Audience for last-impression digital reinforcement. The digital audience and the field list target the same voters through different channels — simultaneous pressure, coordinated timing.
VB reference: Mark Van Buren's R primary win on May 5, 2026 — 43.21% of the vote, 7,065 votes, +695 margin over the nearest competitor — was built on this exact model. $2,500 BMS fee, sub-$1,000 in ad spend.
2. Persuasion Program — Unaffiliated Voters
Scenario: A school board candidate needs to reach unaffiliated voters who are likely to lean conservative but have never voted in a partisan primary.
Platform role: Build the Unaffiliated R-Leaning Persuasion universe using composite signals: property ownership, farm program participation, small business ownership, retail proximity, household party co-residence, and precinct baseline. Generate a persuasion contact list ranked by score. Push the universe to Meta Custom Audiences for Facebook and Instagram targeting.
Why it matters: Unaffiliated voters are invisible to party data products. They do not appear in Republican voter history files. They are exactly the voters most worth reaching in a contested down-ballot race — and the ones most neglected by existing tools.
3. Swing Voter Identification — Contested Township Race
Scenario: A township trustee race in a precinct that has flipped twice in the past four cycles. The candidate needs to know which voters are genuinely persuadable versus which are low-propensity partisans who need motivation, not persuasion.
Platform role: Apply the 5-factor swing voter indicator to separate true swing voters from low-turnout partisans. Swing voters receive persuasion contact (different message, different channel). Low-propensity partisans receive GOTV contact (different message, mail or phone, later in the cycle). Mixing these audiences wastes both contact budget and message clarity.
4. Opponent Broadcast Monitoring — Early Warning
Scenario: A primary race where the opponent is suspected of having more resources. The campaign wants to know when broadcast buys are placed before they hit the airwaves.
Platform role: FCC Political File monitoring flags opponent broadcast activity within 24 hours of contract execution — before the first spot airs. The campaign knows buy size, station selection, and timing before the public does. Resource reallocation and counter-messaging can begin immediately.
5. Agricultural Economy Targeting
Scenario: A county commissioner candidate whose district includes rural townships with significant farm-family populations. Standard voter file segmentation does not distinguish farm families from other rural residents.
Platform role: Cross-reference USDA farm subsidy recipients with the voter file. 604 Licking County voters matched (2.4x Republican baseline lift). Generate a farm-family contact list with parcel acreage data appended. Message on agricultural economy issues — tax policy, drainage board composition, farmland preservation — directly to confirmed farm-family voters.
6. Small Business Owner Outreach
Scenario: A candidate whose primary message is economic development, permitting reform, or tax relief. Needs to reach small business owners specifically.
Platform role: SBA PPP loan records identify 578 Licking County voters as sole-proprietor or small-business borrowers (2.5x Republican baseline lift). Generate a small-business-owner contact list. Push as a Meta Custom Audience for targeted economic-message digital advertising alongside field contact.
7. In-Migration New Registrant Analysis
Scenario: A fast-growing exurban county with significant new voter registrations over the past three years. Are these new registrants reliable Republican votes, or are they transplants from Democratic-leaning urban counties who may complicate the race?
Platform role: Apply IRS county-to-county migration priors to new registrants. Flag registrants from high-Democratic-index origin counties for lower confidence R-lean scores. Flag registrants from high-Republican-index origin counties for elevated confidence. Distinguish genuine swing targets from assumed-but-uncertain Republican additions.
Pricing and Beta Access
Limited Beta — Summer 2026
BMS Voter Intel is entering limited beta in Summer 2026. Capacity is intentionally constrained. The beta cohort will be limited to campaigns where the data can be fully built, validated, and supported before the election.
Current beta priority:
- Ohio — Licking County voter file fully built and validated. Priority engagements for November 4, 2026 general election races.
- Florida — Pipeline engagement pending. Lee County data sourcing underway.
- Illinois — Republican Organization of Wheeling Township engagement; Cook County township races.
Additional counties and states can be onboarded on a build-to-order basis. Timeline and scope depend on data source availability in the target state.
Pricing
Contact for pricing. Engagements are structured by race scope, not by SaaS subscription tier. There is no monthly software fee, no per-seat license, and no party committee approval required.
To inquire about beta access:
Bull Moose Strategy LLC Kelly Fitzgerald, Principal [email protected] bullmoosestrategy.com
Who We Work With
BMS works exclusively with Republican and Republican-leaning independent candidates. We do not take Democratic clients. We do not take clients whose campaigns conflict with the existing client roster. Every engagement is evaluated for conflict before onboarding.
If you are a county commissioner, township trustee, school board candidate, municipal judge, or down-ballot candidate of any type — Republican primary or general election — contact us.
About Bull Moose Strategy
Bull Moose Strategy LLC is a political digital services firm founded in March 2026 and based in Granville, Ohio (Licking County). The firm provides digital advertising, voter intelligence, and campaign infrastructure for Republican and independent down-ballot candidates. Current client engagements span Ohio, Illinois, and Florida pipeline.
Principal: Kelly Fitzgerald — U.S. Marine veteran, OIF 2-2 / GWOT 2004-05. Background in digital strategy, data infrastructure, and small-government policy advocacy.
Current clients and engagements:
- Mark Van Buren — Licking County Commissioner, R primary winner May 5, 2026 (43.21% / 7,065 votes / +695 over nearest competitor). November 4, 2026 general election campaign active.
- Republican Organization of Wheeling Township — Cook County, Illinois township GOP committee. Mailchimp, digital infrastructure, and voter data pipeline.
- Amanda Cochran — Lee County (FL) Commissioner District 5 candidate. R primary August 18, 2026. Prospect engagement active.
Entity: Ohio LLC, registered March 19, 2026. EIN 41-4999862. Registered Agent: Registered Agents Inc.
Brand: "Candidates, Not Parties."
The firm is named for Theodore Roosevelt's Progressive Party — the Bull Moose — not as an ideological statement, but as a reminder that the best candidates run on competence and conviction, not institutional protection.
BMS Voter Intel is a Bull Moose Strategy proprietary product. All 18 data sources are public record. Universe construction methodology, scoring weights, swing indicator formulas, and universe SQL are proprietary and not disclosed. This whitepaper describes platform capabilities, data sources, and use cases at a high level for prospective client evaluation. No specific pricing, outcomes, or performance guarantees are made.
For questions about methodology, data sourcing, or compliance posture, contact [email protected].
Bull Moose Strategy LLC — Granville, Ohio — bullmoosestrategy.com
Version 1.0 — May 2026