Voter intelligence and targeting

Culper is built on the county's public master voter roll — the authoritative record of who is registered and how they have actually voted — not a stale commercial snapshot. Where Culper operates today, that roll is loaded and kept current, so your universe matches the county's as registrations, moves, and cancellations flow through.

Alongside it, public county CAD (appraisal district) records supply authoritative addresses and parcel coordinates. That is what makes a block-walk route land on the right door and a mail list dedupe cleanly by household. Both are public records, and we are expanding coverage to every county and parish before the next cycle.

• County master voter roll — the source of truth for registration and real vote history
• Public county CAD / parcel data — authoritative addresses and lat/long for mapping and walks
• Every voting record you see comes straight from the county roll, never a modeled guess

During early voting and on election day, Culper ingests the county's cumulative early-vote and turnout files as they post. Your lists update to show who has already voted, so walkers, texters, and callers stop spending touches on banked voters and shift to the people who still need a nudge.

The same feed powers a live turnout view: you can watch actual votes come in against your model and re-target the chase in the hours that matter, instead of finding out the next morning.

Have a licensed extract from L2, i360, or Aristotle? Import it and Culper resolves it into the record — newest data wins, field by field, with the provenance of every value kept. There is no vendor lock-in and no hand-mapping two hundred columns: the importer auto-detects common formats and maps them for you.

The same path takes an NGP VAN export or any spreadsheet. A raw file becomes a resolved, queryable, tag-able record — the flow shown in the animation on the home page.

Culper normalizes every vendor's columns into one consistent set of smart attributes — party, age, turnout history, issue signals, homeowner status, and more — so a query behaves the same no matter where the data came from. You stop reconciling three vendors' column names and just ask the question.

On top of that, your campaign creates its own tags — "Wants a sign," "Volunteer," "Persuadable" — that travel with the voter across every feature: walks, texts, events, and lists. Feedback from the field writes tags and answers straight back onto the record.

Type what you want — "Republicans over 55 in precinct 42 who voted the last three primaries" — and get the list. No analyst, no SQL, no query builder to learn. The natural-language layer turns your ask into a safe, constrained query that runs only against your own campaign's data, then hands back a saved, walkable, textable segment.

Structured filters, saved and pinned lists, and select-all-and-materialize are there too, for when you would rather build a segment by hand.

Culper turns your universe into a plan: how many votes you need to win, which voters are worth a touch, and in what order. Score voters by persuadability and turnout, carve them into walk and outreach universes, and watch the math to your win number as supporters get contacted and bank their votes.

It reframes the whole campaign from "how many doors did we knock" to "how close are we to the number that wins."

• Your win number, and the exact universe that gets you there
• Persuasion and turnout universes, carved and prioritized
• Progress to goal that updates as contacts and early votes land

This is what sets Culper apart. For each voter, Culper keeps a running count of every time your campaign reached them and how — walks, SMS, peer-to-peer, ads, yard signs, mail, and email — so the record reads, in plain terms, "reached six times: two walks, one text, one P2P, one mailer, one email." A contact heatmap rolls the same data up to the precinct so you can see coverage at a glance.

Next to the count sits a probability the voter will actually turn out, blending real vote history with model signals. Together they answer the question every campaign is really asking: who has heard enough from us, and who hasn't heard at all.

• Per-voter contact history across all seven channels
• A turnout probability for every voter
• A precinct-level contact heatmap for coverage at a glance

The flip side of touch-point tracking is the most valuable view of all: the key voters who are not hearing your message. High-value supporters and persuadable voters with zero or too-few touches rise to the top, so you can fill the gaps before election day.

That is the difference between a universe you contacted and a universe you actually moved — and it is usually where a close race is won or lost.

Once the county posts final turnout, Culper compares who you expected to vote against who actually did. Every voter gets a verdict — expected and voted, expected and didn't, or turned out when the model said they wouldn't — read alongside the touches they received and the likely reason for a miss.

It is the debrief that makes the next race sharper: which universes performed, which channels moved turnout, and where the model was wrong. Deeper post-election modeling is an active area we are building out cycle over cycle.