Rule-Based Swing Voter Identification

The Problem

"Swing voter" is one of the most overused and least-defined terms in political data. Vendors sell swing voter scores as a black box — a number between 0 and 100 that supposedly reflects persuadability, generated by a model that the campaign cannot inspect, cannot explain, and cannot reproduce. When you ask what goes into the score, the answer is usually some version of "proprietary methodology." The problem is not that modeling is wrong. The problem is that an opaque score gives you no way to evaluate whether it is working, no way to explain your targeting rationale to a skeptical campaign manager, and no way to audit your own results after the election.

What BMS Voter Intel Does

BMS Voter Intel identifies swing voters through a transparent, rule-based classification system built on five explicit, verifiable factors. Every voter who lands in the swing universe got there because they meet a defined threshold on defined criteria — criteria that a campaign manager can read, understand, challenge, and replicate independently. There is no black box. If you ask why a specific voter is classified as swing, there is a specific answer.

How It Works (Without Revealing IP)

The five-factor rule set draws on public voter file attributes: registration party relative to household composition, cross-party voting history in available election cycles, registration recency and change history, geographic location relative to partisan precinct context, and enrichment signals that indicate cross-pressured household characteristics. A voter must satisfy a combination of these factors to enter the swing classification — not just one. The thresholds are set to be conservative, meaning the swing universe will be smaller and higher-confidence than a broad model that assigns every voter some probability of being persuadable. That tradeoff is intentional. A small, high-confidence swing list is more actionable than a large, speculative one.

Real Numbers from the Van Buren Campaign

Applied to 124,614 Licking County registered voters, the five-factor rule set identified exactly 106 swing voters — 0.085% of the total electorate. That number is not a modeling artifact. It is a validated finding: Licking County is not a persuasion county. It is a GOTV county. Nearly every competitive political move available to a Republican candidate in Licking County is a turnout move, not a message move. The 106 swing voters that the rule set surfaced are genuinely cross-pressured — not merely lower-propensity Republicans who needed a different kind of classification. Knowing this with confidence shaped the entire campaign's strategic resource allocation. The campaign invested in turnout infrastructure, not persuasion mail, and won the primary by 695 votes.

What This Means for Your Race

In some counties, 5% of the electorate is genuinely persuadable — voters who have split-ticket histories, household-level cross-pressures, and no strong partisan anchor. In others, it is less than 0.1%. Knowing which situation you are in before you finalize your budget is the difference between an efficient campaign and one that spends 30% of its mail budget trying to persuade voters who were never going to move. A transparent rule-based system also gives you something a black-box score never can: the ability to defend your targeting decisions to your candidate, your donor, and yourself.

Get Beta Access

BMS Voter Intel is in limited beta for Summer 2026 — Republican and independent down-ballot campaigns in Ohio, Florida, and Illinois. Contact [email protected] or visit bullmoosestrategy.com/voter-intel to learn how swing classification applies to your specific county and race type.