Strategy
Who Actually Lives Behind That Door? Why Precinct-Level Data Misses the Household
A typical canvass shift covers 40-60 doors in four hours. Decades of field experiments from Yale's Institution for Social and Policy Studies show that personal canvassing lifts turnout by an average of 6 percentage points. But that 6% is an average across successful contacts with the right voters. Most walk lists are built from precinct-level data that tells you about the neighborhood—not the household standing in the doorway.
The Precinct Problem
A typical precinct contains between 500 and 3,000 registered voters. Precinct-level demographic data—the kind pulled from the Census Bureau's American Community Survey —describes the statistical average of a geographic area. It does not describe the person answering the door at 7:15 PM on a Tuesday.
The ACS publishes estimates at the block-group level, which covers roughly 600 to 3,000 people. Even at that resolution, a "block group" that is 70% homeowners, median age 52, with 14% Latino population is a statistical composite. The actual household at 1214 Maple Street may be a 29-year-old renter living in her parents' house, a multi-generational family with three distinct voting histories, or a recently naturalized citizen with no prior registration at all.
What Precinct Data Tells You vs. What You Actually Need to Know
| Precinct-Level Data Point | Household Reality | Targeting Implication |
|---|---|---|
| Median age: 58 | Adult child (age 24) lives with aging parents | GOTV message aimed at seniors misses the low-propensity young voter at the door |
| Owner-occupied: 74% | Owner lists their adult daughter as resident; she rents from them informally | Homeownership-based messaging lands wrong |
| Turnout rate: 68% (2022) | Individual voter has voted once in six elections | Precinct average masks a low-propensity voter who needs aggressive GOTV, not a soft reminder |
| Party registration: 62% Democrat | Registered Democrat who votes in Republican primaries | Registration without vote history creates false confidence in support score |
| Income: median $72K | Household includes a 2018 registrant with zero vote history | Economic messaging designed for established voters may not resonate with a new registrant |
Sources: Census Bureau ACS Methodology, Yale ISPS GOTV Field Experiments
What's Actually Behind the Door
Multi-Generational Households
Pew Research Center data shows that roughly 20% of Americans—approximately 66 million people—live in multi-generational households. This share has more than doubled since 1971 and has accelerated since 2020. A precinct marked as "homeowner, 65+" may reliably produce a 68-year-old homeowner at 9 of 10 doors—and a 22-year-old grandchild at the tenth, a voter who has never cast a ballot and whose turnout propensity looks nothing like the block-group average.
Multi-generational households also contain voters with divergent partisan identities. A GOTV knock designed around one generation's top issue (Medicare, Social Security) delivered to another generation (student debt, housing costs) generates friction rather than enthusiasm.
Homeowner vs. Renter Confusion
Property records identify legal ownership. They do not identify residency. Young adults living in parent-owned homes, informal rentals, and short-term arrangements create systematic mismatch between ownership data and who actually answers the door. A canvasser armed with talking points about property tax policy has made a confident assumption that may be wrong 20-30% of the time, depending on the neighborhood.
Presidential-Only Voters
Approximately 40% of registered voters participate only in presidential general elections. Census Bureau Current Population Survey data consistently shows that midterm turnout runs 15-20 percentage points below presidential turnout for the same registered voter population. Knocking a presidential-only voter's door during a state legislative primary is wasted effort—they are not in the electorate for that race regardless of their preferences.
Household Types Your Walk List Probably Ignores
| Household Type | Why It Matters | What Precinct Data Shows Instead |
|---|---|---|
| Multi-generational | Multiple voters with different ages, histories, and issue priorities sharing one address | Single demographic profile based on block-group averages |
| Split-partisan | Registered voters in the same household with opposing party registrations | Precinct-level party breakdown (e.g., "60% Democrat") |
| Renter in owner-occupied home | Young adult or informal tenant whose actual concerns differ from "homeowner" messaging | Ownership rate (% of housing units that are owner-occupied) |
| New mover | Recently registered at address; no vote history at this location; may not know local candidates | Same voter file record as a 20-year resident |
| Presidential-only voter | Reliably off-year absent; deploying field resources on them for a midterm is waste | Registered voter in a "high-turnout" precinct |
| Recently naturalized citizen | First-cycle voter with no prior history; extremely high potential if contacted correctly | Same registration date treatment as a lapsed voter who re-registered |
Sources: Pew Research Center, Multi-Generational Living (2022), Census Bureau Voting and Registration Data
The Contact Rate Problem
Even a well-built walk list runs into a structural reality: most people are not home, do not answer the door, or are not the target voter when they do answer. Industry contact rates for door-to-door canvassing average 20-30%. The Center for Campaign Innovation's 2024 post-election survey found that only 19% of rural voters reported a campaign contact at their door, compared to 34% of urban voters—suggesting contact rates vary dramatically by geography and are often lower than campaigns report internally.
The math on a naive canvass shift is sobering. A canvasser who walks 60 doors in a four-hour shift:
- Contacts made: 20-30% contact rate yields 12-18 conversations from 60 attempts
- Actual target voters contacted: Of those contacts, some will be wrong address, non-targets, or already strong opposers—reducing real targets to perhaps 8-12
- Turnout lift generated: A 6% lift applied to 8-10 genuine targets produces less than one additional vote per shift
- The implication: If contact rate improves and targeting precision improves, the same shift generates meaningfully more votes per hour than a precinct-average list
Naive vs. Targeted Canvassing: Per-Shift Math
| Metric | Naive (Precinct-Level List) | Targeted (Household-Level List) |
|---|---|---|
| Doors attempted per shift | 60 | 45 (tighter geography, more travel time per door) |
| Contact rate | ~25% | ~30% (clustered routing improves contacts) |
| Contacts made | 15 | 13-14 |
| % of contacts who are real targets | ~50% (wrong household member, non-targets) | ~75% (household-matched to specific voter) |
| Real target contacts | ~8 | ~10 |
| Votes generated (6% lift) | ~0.5 | ~0.6 |
| Volunteer hours per vote | ~8 hours | ~6.7 hours |
Contact rate and targeting estimates based on: Yale ISPS GOTV Experiments and Center for Campaign Innovation, 2024 Survey
What Household-Level Intelligence Actually Looks Like
Moving from precinct averages to household-level intelligence requires assembling data at the individual and address level, then reasoning about who lives together and what each person's history looks like.
- Household clustering: Identifying all registered voters at a single address, their relationships (surnames, ages, registration dates), and likely head of household—so a canvasser knows who to ask for and what household composition to expect
- Individual voter history: Not registration status, but participation: which elections did this specific person vote in? General-only voters, primary voters, off-year voters, and new registrants each have distinct contact strategies
- Property matching: Comparing the voter file surname and age against property ownership records to distinguish true homeowners from adult children in parent-owned homes or informal rentals
- Individual turnout modeling: Probability scores at the person level, not precinct averages—built from vote history, age, registration length, geography, and demographic priors
- Per-door talking points: Walk list entries that surface household composition and individual context, so a canvasser arriving at a multi-generational household knows to ask for a specific person and lead with the right issues
How Data Quality Affects Field ROI
Yale ISPS research estimates the cost per vote generated through door-to-door canvassing at roughly $20-$50 per vote, depending on contact rates and local labor costs. That estimate assumes the canvasser is talking to actual target voters. When contact quality degrades—because the list is built on precinct averages rather than household intelligence—the true cost per vote rises significantly.
$100K Field Investment: Precinct List vs. Household List
| Metric | Precinct-Level List | Household-Level List |
|---|---|---|
| Total volunteer/staff hours purchased | ~5,000 hours | ~5,000 hours |
| Doors knocked | ~75,000 | ~60,000 (tighter routing) |
| Real target voter contacts | ~9,000 (contact rate 25%, target hit rate 50%) | ~13,500 (contact rate 30%, target hit rate 75%) |
| Estimated votes generated (6% lift) | ~540 | ~810 |
| Cost per vote | ~$185 | ~$123 |
| Wasted contact attempts | ~66,000 doors (no contact or wrong target) | ~45,000 doors |
The household-level list generates roughly 50% more votes from the same budget—not by working harder, but by directing the same effort toward doors that actually matter.
Cost-per-vote range based on: Gerber & Green, Yale ISPS, "The Effects of Canvassing, Telephone Calls, and Direct Mail on Voter Mobilization"
The Canvasser Experience Problem
Bad data does not just waste money. It degrades the human infrastructure that makes field operations possible.
Canvassers who knock doors and have real conversations stay. Canvassers who knock wrong addresses, encounter hostility because they are pitching the wrong message to the wrong household, or spend the shift walking past closed doors that never open—quit. The Center for Campaign Innovation's 2024 survey found that volunteer retention is one of the top operational challenges campaigns report, with a significant portion citing poor data quality and low-quality contacts as contributing to attrition.
An experienced canvasser who has had ten meaningful conversations in a shift returns. They train new volunteers. They bring enthusiasm into the next shift. An inexperienced canvasser who knocked forty doors and spoke to six people about topics they did not care about does not come back.
- Volunteer retention is compounding: A retained canvasser from week one becomes a trainer by week four. A lost canvasser requires recruiting, onboarding, and training a replacement—at roughly the same cost as deploying three fully productive shifts
- Canvasser confidence drives contact quality: Volunteers who know why they are knocking a specific door—who lives there, what they care about, what their vote history looks like—deliver better conversations than volunteers handed a street address and a generic script
- Bad data creates hostile contacts: Knocking a confirmed Republican's door with a Democratic GOTV pitch or showing up at a vacant address does not just fail—it occasionally generates the kind of negative interaction that discourages volunteers from returning
- Most political professionals believe data quality directly affects outcomes: Yet far fewer report confidence in the accuracy of their voter data, per the CCI 2024 post-election survey. The confidence gap is largest precisely where field operations are most expensive to run
Give your canvassers doors worth knocking.
Walk lists built on household intelligence mean every volunteer conversation lands with the right person, the right message, and the right level of urgency—driving both contact quality and volunteer retention.
Build smarter walk lists →The Bottom Line
The 6% turnout lift from canvassing is real. It has been replicated across dozens of field experiments over three decades of rigorous research. But it is an average that hides enormous variation—between campaigns that deploy their field resources against the right voters and campaigns that deploy them against geographic averages.
Precinct-level data is not wrong. It is a useful starting point for understanding neighborhoods. But it is a poor proxy for the household behind each door: whether a young adult or an aging parent answers, whether the resident is a homeowner or an informal tenant, whether the registered voter at this address has ever shown up for an off-year election or consistently votes only when a presidential race is on the ballot.
Campaigns that invest in household-level intelligence—combining voter history at the individual level, property matching, household composition modeling, and individual turnout scores—concentrate the 6% lift on the voters who matter. Campaigns that run precinct-level lists spread that lift thin across doors that won't answer, voters who weren't going to show up regardless, and households that received the wrong message entirely.
The ground game is not about how many doors you knock. It is about how many of the right conversations you have with the right people. The difference is the data behind the list.
Filed under strategy on March 28, 2026.