A panel read is only as good as what anchors it. Attach the data sets your campaign has — voter file cuts, election returns, polling, field notes — and the panel reads your message against them instead of guessing.
Attach what your campaign has. Each data set adds a grounding requirement to the run; you paste the actual numbers or notes in step 2 of the builder. Attach any combination, then start the run.
Voter-file cuts, modeled scores, turnout history, contact history, and district geography.
Preserves the real target universe instead of making the panel infer party mix, geography, turnout, or contact history from message text.
district · precinct · party / modeled score · turnout · contact history
Use aggregate or permissioned campaign data only. Do not paste private voter identities into the demo.
Modeled partisanship, issue scores, demographic priors, vote propensity, and persuasion universes.
Lets the run distinguish base, persuasion, primary, low-propensity, and issue-specific audiences without collapsing them into one generic voter group.
modeled party · vote propensity · issue scores · segments · geography
Treat modeled scores as priors, not facts about individual voters.
Statewide, county, district, precinct, and runoff baselines from election officials or vetted public datasets.
Grounds the message read in real partisan history and recent turnout context before a campaign uploads proprietary files. Coldread includes official 2024 presidential statewide totals for all 50 states + DC; add county, district, precinct, or voter-file cuts when the read needs local precision.
vote totals · margins · turnout · precinct · year
Past returns explain context. They are not a turnout or persuasion forecast.
ZIP/ZCTA, county, district, and population-growth context from ACS and Census population estimates.
Shows whether the audience is stable, fast-growing, exurban, urban, older, younger, renter-heavy, owner-heavy, or otherwise demographically distinct.
population · growth · age · income · education · race / ethnicity
Aggregate demographics are recruitment context, not voter-file counts or individual preferences.
Registration, voting method, provisional ballot, mail ballot, and election-administration friction.
Improves GOTV, deadline, vote-method, and voting-process copy by accounting for registration and ballot-friction patterns.
registration · mail voting · provisional ballots · turnout · administration
Administration data explains process friction. It does not say how voters feel about the message.
Committee, contribution, spending, donor geography, amount band, and repeat-giving context.
Supports fundraising reads, opponent-money proof, small-dollar donor geography, and PAC or committee language. State races file with state ethics commissions (e.g., Missouri MEC), not the FEC.
committees · receipts · disbursements · donor geography · amount bands
Public finance records are not donor intent and should not stand in for the whole electorate.
Missouri State Auditor race context, candidate committee facts, public campaign claims, election dates, and state-race finance rules.
Lets a Quentin Wilson run test the actual Auditor persuasion burden: voters must understand the office, believe the watchdog/resume frame, and see why one statewide check matters in an R+18 state.
MEC committee · SOS calendar · candidate bio · office stakes · Missouri returns
Use the public race kit as a starter brief. Campaign-owned voter-file, polling, field, donor, and event data should override public proxies.
Audit reports, state spending, public-corruption/fraud context, budget transparency, and measurable-government proof points.
Turns an Auditor message from abstract good-government language into testable proof: which waste, audit, spending, or performance example makes voters believe the office matters.
audit reports · state spending · fraud / waste · budget data · performance claims
Audit and spending data can support accountability claims, but the message must not imply misconduct without a sourced report or campaign-approved fact base.
County, metro, and state employment, unemployment, income, and local economic trend context.
Makes jobs, wages, cost-of-living, and economic competence messages answer the local reality instead of leaning on national talking points.
unemployment · labor force · personal income · GDP · county / metro
Economic indicators explain conditions. They do not prove which candidate caused them.
Rent burden, owner cost burden, overcrowding, tenure, income bands, and local housing-needs data.
Grounds housing, rent, property-tax, growth, and affordability messages in local pressure instead of generic cost copy.
rent burden · owner burden · tenure · income band · overcrowding
Housing burden is a condition measure. Pair it with local geography before using it to define an audience.
School district, enrollment, staffing, fiscal, and public education context for education and family messages.
Helps school, teacher, parent, growth, and tax messages reflect the local education system voters recognize.
district · school · enrollment · staffing · finance · grade span
School-system data is institutional context. Do not infer parent opinion from enrollment counts alone.
County, place, tract, and ZCTA health outcomes, risk behaviors, access measures, disability, and social-needs data.
Grounds healthcare, public safety, addiction, disability, elder-care, and quality-of-life messages in local conditions.
health outcomes · access · risk behavior · disability · ZCTA / county
Health indicators describe community conditions, not individual diagnoses or voter intent.
County metro/nonmetro classification, adjacency, urbanization, and rural/suburban context.
Prevents place-blind reads by distinguishing rural counties, exurbs, small metros, inner suburbs, and large metro counties.
metro · nonmetro · adjacency · urbanization · county
Place type should shape context and language, not substitute for voter-file or precinct data.
Agency, county, city, and national crime, victimization, and public-safety trend context.
Keeps public-safety messages grounded in reported local conditions and separates official crime trends from voter fear, anecdote, or opponent framing.
agency · county / city · offense · victimization · trend year
Reported crime data and victimization surveys measure different things. Name the geography, year, and data caveats.
Traffic volume, commute, road safety, crash fatality, and transportation-system context.
Makes growth, infrastructure, commute, gas-tax, pedestrian-safety, and road-improvement messages answer the daily friction voters notice.
traffic volume · fatal crashes · commute · road class · state / county
Traffic and crash data show conditions and risk. They do not identify who voters blame without polling or field notes.
Federally declared disasters, severe weather, billion-dollar disasters, assistance eligibility, and local resilience context.
Grounds climate, insurance, emergency management, flooding, wildfire, grid, and recovery messages in specific local risk and recent events.
disaster type · declaration · county · assistance · event cost · year
Disaster data can show exposure and history. It is not a forecast for a specific household or precinct.
Electricity prices, generation mix, utility-sector trends, reliability context, and state energy conditions.
Helps power-bill, clean-energy, grid, utility, and cost-of-living messages reflect actual state and regional energy conditions.
electricity price · generation · utility · sector · state · trend year
Energy data explains costs and supply mix. It does not prove voter attribution or policy causality alone.
Broadband availability, serviceable locations, subscription, device access, and internet-use context.
Grounds rural broadband, homework gap, telehealth, small-business, and digital government messages in actual coverage and adoption gaps.
broadband availability · location · provider · internet use · device access
Availability is not the same as affordability, adoption, speed experience, or service quality.
Farm production, acreage, commodity, county typology, manufacturing, recreation, and rural economic dependence context.
Makes farm, food, rural hospital, small-town economy, trade, and land-use messages speak to the actual local economic base.
commodity · acreage · farm count · county typology · economic dependence
County economic dependence and farm data are place context, not an automatic stand-in for partisan identity.
Food access, food insecurity proxies, SNAP participation, store access, and local food-environment context.
Improves cost-of-living, child hunger, grocery access, farm bill, and benefits messages by grounding them in household pressure and place access.
SNAP · store access · low income · food environment · county
Program participation and food access data should be handled as aggregate conditions, not labels for individual voters.
Language spoken at home, English proficiency, nativity, citizenship, immigration-flow, and community-composition context.
Helps outreach, translation, immigration, workforce, schools, and community-safety messages account for language access and immigrant-community realities.
language · English proficiency · nativity · citizenship · state / county
Use aggregate language and immigration context for access and framing. Do not infer status, eligibility, or opinion for individuals.
Veteran population, age, disability, income, VA expenditures, benefit categories, facilities, and military-community context.
Grounds veterans, defense, VA, health care, jobs, and base-community messages in the size and needs of the local service community.
veteran population · disability · income · VA expenditures · county / state
Veteran population and VA spending data do not imply veteran opinion. Pair with polling or field notes for message preference.
Poll toplines, demographic crosstabs, issue salience, favorability, ballot, and message-test data.
Gives the read a known opinion baseline and highlights which groups already show issue demand, proof gaps, or trust risk.
toplines · crosstabs · issue salience · favorability · ballot
Include sample size, field dates, mode, and margin notes when available.
Door notes, phone-bank patterns, organizer feedback, recurring objections, and local proof gaps.
Makes the simulated voters react to the real objections and language organizers are hearing instead of generic national talking points.
objections · local language · proof gaps · issue priority · volunteer notes
Summarize themes and anonymize individual voter notes.
Survey, event, signup, reply, and concern-intake themes from the campaign's own listening loops.
Makes a message react to what Missourians are actually telling Quentin: what works, what does not, what worries them, and which outcomes they want government to improve first.
survey themes · event notes · issue priority · worry language · local proof gaps
Summarize and anonymize responses. Do not paste private contact details, raw phone numbers, or identifiable voter notes into a demo run.
Donor segments, supporter history, email engagement, volunteer cohorts, and contribution patterns.
Improves fundraising, mobilization, volunteer, and base-activation reads by separating donors from persuadable voters.
donor tier · engagement · volunteer history · supporter segment · geography
Do not use supporter or donor data as a proxy for persuadable voters.
Local press clips, opposition research, fact checks, candidate biography, endorsements, and voting records.
Lets the run test whether claims feel supported, overreaching, locally credible, or vulnerable to rebuttal.
clips · quotes · records · endorsements · oppo claims
Separate sourced facts from claims the message wants voters to accept.
Replies, comments, DMs, search trends, creator discourse, and issue-language patterns.
Shows live phrasing, sarcasm, misinformation hooks, emotional triggers, and terms the campaign should avoid or reuse.
comments · replies · DM themes · search terms · misinformation hooks
Social volume is not representative opinion. Use it as language and objection discovery.
The default voter-data baseline used before additional campaign data is supplied.
Provides broad message-risk coverage, voice diversity, and a baseline read when campaign data is incomplete.
age · region · occupation · values · media diet · voice
This is simulated baseline research. Mark campaign reads as provisional when no real source data is attached.
Real certified 2024 returns for all 50 states + DC ship with Coldread — vote totals, statewide margin, and whether the state registers voters by party. Pick a state to see its numbers, then attach them to your next run as source data.
Select a state to preview its 2024 baseline — the exact text a run receives as grounding.
Statewide election-environment context, not a local quota or a forecast. Supplied voter-file or registration counts always take precedence.