Guide 02

Political Canvassing and GOTV: The Field Program Guide

A campaign field program is a machine for converting a voter file into votes, and it has exactly one hard constraint: the deadline does not move. Everything difficult about political canvassing follows from that. You cannot extend the season, you cannot ship late, and a plan that would have worked with three more weeks is simply a plan that failed.

This guide walks the whole cycle. Where the universe comes from, how precincts get cut, how you recruit and keep volunteers who are not being paid, the switch from persuasion to turnout, and what the final week actually looks like when it is being run well. It assumes you know what canvassing is; if you do not, start with the complete guide to door-to-door canvassing and come back.

What political canvassing is

Political canvassing is door-to-door voter contact run from a voter file. A canvasser works an assigned list of registered voters, has a short conversation about a candidate or a measure, and records a disposition. The conversation is the visible product. The disposition is what the campaign spends later.

That second point is the thing distinguishing a field program from volunteers knocking on doors. A campaign that talks to twenty thousand voters and records nothing has bought twenty thousand conversations and no list. A campaign that talks to eight thousand and records every one has a turnout universe. In a close race the second campaign wins, and it is not close.

The voter file and the universe

Everything starts with the file: the public record of registered voters, usually purchased or accessed through a party or vendor, carrying registration status, party where the state records it, address, and vote history. Vote history is the most valuable column in the file and the most misread. It records whether someone voted, never how.

From the file you cut a universe using two axes: likelihood of supporting you and likelihood of voting. Those two produce four groups, and only two of them are worth doors. High-support high-turnout voters will vote for you anyway and mostly need leaving alone. Low-support high-turnout voters are your opponent's, and knocking them raises their salience, which is a genuine risk rather than a theoretical one. The doors worth walking are high-support low-turnout, who are your GOTV universe, and the genuinely undecided, who are your persuasion universe.

Getting this wrong is the most expensive mistake in field, because it wastes the whole season rather than one shift. It is also the mistake that feels most productive while you are making it, since knocking everybody produces excellent door counts.

The other half of the work is hygiene: deduplicating, standardizing addresses, resolving apartment numbers, and reconciling the file against what canvassers report from the doors. People move constantly, and a file that is a year old will send canvassers to a meaningful number of addresses where the target no longer lives. Those moved records need to come back out of the universe, not sit in it inflating your coverage denominator.

How campaigns organize the field

The standard structure exists because it has survived a lot of cycles. A field director owns the program and the number. Field organizers own a geography and the volunteers in it. Volunteer team leaders run shifts. Canvassers walk turf. On a small race one person is several of these, but the roles do not disappear, they merge.

The unit of work is the shift, not the day. A shift is a launch, a turf assignment, three or so hours of walking, and a debrief where the packets come back and the data goes in. Campaigns that skip the debrief lose data and lose volunteers, in that order.

Turf gets cut along precinct lines where possible, because precincts are how election results arrive and because matching your field geography to your results geography is the only way to learn anything after the fact. If your turf boundaries cut across precincts, you will never be able to answer whether the canvass moved the number.

Recruiting and keeping volunteers

Field programs run on unpaid labor that can stop showing up at any time and suffers no consequence for it. This one fact drives most of what is strange about campaign management.

The arithmetic every campaign eventually learns the hard way is that a confirmed commitment is not an attendance. A meaningful share of people who say yes will not appear, and the gap widens with weather, with how far ahead the commitment was made, and with whether anyone called to confirm. Programs that plan on raw sign-up numbers are chronically short-staffed on Saturday morning, and the ones that plan for the gap and confirm the day before are not.

Retention beats recruitment on cost every time. A volunteer on their fourth shift knocks faster, records better data, and can be handed a difficult turf. Losing them and replacing them costs a full training cycle. The practices that retain are small and specific: give people the same turf so they build familiarity, tell them what the data showed, put them in pairs, and have someone who knows their name greet them at launch.

Persuasion, and the voter ID

Persuasion canvassing has two jobs and campaigns forget the second one. The first is to move the voter. The second, which is the one that pays off later, is to find out where they stand and write it down.

A voter ID is a recorded assessment of a voter's lean, usually on a short scale. It is captured during a persuasion conversation, and it is the raw material of the entire GOTV program. A campaign that runs persuasion without capturing IDs has spent the expensive part of the season and produced nothing that the cheap part can use.

The conversation itself works better when it is a conversation. The field research that has held up best points the same direction as experienced organizers do: the canvasser who asks an open question, listens, and responds to what the person actually said moves voters, and the one who delivers a message does not. That is hard to script and harder to train, which is why persuasion programs need more training investment per canvasser than turnout programs do.

The switch to GOTV

At some point the campaign stops trying to change minds and starts trying to produce ballots. This is a discrete switch, not a gradual fade, and it changes the universe, the script, the shift length, and the definition of a good door.

The GOTV universe is built from your IDs: supporters whose turnout history says they might not vote. The script gets shorter and more concrete, because the job is no longer persuasion but logistics. Where is your polling place, when are you going, do you know it moved, do you need a ride. Asking a supporter to commit to a specific time is measurably more effective than asking whether they plan to vote, and it costs nothing to do.

The operational mistake here is running a GOTV script at a persuasion universe because nobody rebuilt the lists. Reminding an opponent's voter to vote is worse than not knocking them at all.

The final week

The last several days are a different operation. Shifts run longer and start earlier, turf gets recut daily rather than weekly, and the reporting cadence goes from daily to hourly. The value of a knock rises steeply as election day approaches and then falls off a cliff the moment polls close, which is why a campaign that is behind on coverage in the final week cannot fix it by working harder in the following one.

What separates a good final week from a frantic one is that the good one was planned in September. The universes are already cut, the packets already built, the volunteer shifts already confirmed, and the only decision being made each morning is where to move people based on last night's coverage. Campaigns improvising this in the last week burn their field organizers and under-cover their best turf.

Where early and mail voting exist, the file changes underneath you daily: voters who have already returned a ballot should come off the turnout universe immediately, because knocking them wastes a contact and irritates a supporter. That reconciliation needs to be a standing daily task with an owner, not something someone remembers to do.

Ballot measures and issue advocacy

Issue campaigns look like candidate campaigns and behave differently in ways that catch experienced staff out. There is no candidate to like, party cues are weak or absent, and voters frequently have no prior opinion at all, which makes persuasion more available and the conversation longer.

Ballot measure canvassing also has to contend with ballot language that is often confusing by construction, so a large share of the doorstep conversation is spent establishing what a yes vote even does. Petition drives to qualify a measure are a different job again, with a signature threshold and a filing deadline where validity rate matters as much as raw count.

Paid canvassers versus volunteers

Most programs of any size end up running both, and they behave differently enough that managing them identically is a mistake.

Volunteers are cheap, genuinely motivated, and unreliable in aggregate. They cancel, they arrive late, and they cannot be scheduled with any confidence more than a few days out. What they bring is authenticity: a neighbor who believes what they are saying has a different conversation than someone on an hourly rate, and in persuasion work that gap shows up in the results.

Paid canvassers are predictable, schedulable, and trainable to a standard, which is why any program with a hard coverage target eventually hires some. The cost is that compensation shapes behavior, and it shapes it toward whatever you measure. Paying by the door produces a lot of doors and short conversations. Paying by the hour with no quality check produces slow shifts. Paying by ID produces IDs that are not real.

That last one is the failure mode to design against. Falsification is rare but not hypothetical, and it is corrosive because a fabricated ID does not just waste one door, it poisons the GOTV universe you build from it. The defenses are unglamorous and they work: GPS on the recorded knock, spot-check callbacks on a sample of reported contacts, and watching for canvassers whose numbers are implausibly good rather than only praising them. A program that has never audited its own dispositions does not know whether its data is real.

The practical split most campaigns land on is paid staff for the coverage backbone, with volunteers concentrated where authenticity pays: their own precincts, their own networks, and the final GOTV weekend when enthusiasm is high and the job is simple.

The coverage math nobody does early enough

Do this arithmetic in the first week of the program, not the last.

1. Take the universe size.
2. Divide by attempts per canvasser shift to get shifts required for one full pass.
3. Multiply by the number of passes you want, remembering that a not-home is not a contact and most universes need two or three passes to reach a decent share of it.
4. Divide by shifts remaining on the calendar to get canvassers needed per shift.
5. Inflate that by your no-show rate to get the number you actually have to recruit.

Campaigns that run this in week one discover in week one that the universe is too big, and cut it while cutting it is still cheap. Campaigns that run it in October discover the same fact when the only remaining option is to abandon turf.

What happens to the data afterwards

Election night ends the campaign and, in most organizations, ends any thought about the field data. This is the cheapest available mistake, because the file you just spent a season building is the single most valuable thing the campaign produced other than the result.

Three things are worth doing in the week after, while anyone still remembers.

Reconcile against the result. If turf stayed inside precinct lines, you can now compare your coverage and your IDs against what each precinct actually did. That comparison is the only honest feedback a field program ever gets, and it will usually surprise you. Precincts where the canvass was heavy and the result underperformed are worth understanding rather than explaining away.

Write down what the file learned. Every moved record, every wrong address, every apartment that turned out not to exist is a correction the next campaign will otherwise rediscover at its own expense. So is every ID. Vote history will tell a future campaign that someone voted; only your file will tell them why.

Keep the volunteers. The list of people who actually showed up and knocked is a genuinely scarce asset, and it degrades fast. A campaign that thanks its canvassers properly, tells them what the data showed, and keeps the list somewhere findable starts the next cycle with a recruitment problem that is already half solved. Most campaigns dissolve and lose it entirely.

There are real constraints on what can be shared and retained, and they vary by jurisdiction and by the terms attached to whatever voter file you licensed, so check before transferring anything to another committee or organization.

Common questions

What is political canvassing?

Door-to-door voter contact run from a voter file. Canvassers work an assigned list of registered voters, have a short conversation about a candidate or issue, and record a disposition. The record is what the campaign uses later to decide who to turn out.

What is the difference between persuasion and GOTV canvassing?

Persuasion targets undecided or soft voters and tries to change how they vote. GOTV targets voters already identified as supportive and tries only to make sure they cast a ballot. Different universes, different scripts, different points in the calendar.

What is a voter ID and why does it matter?

A recorded assessment of how a specific voter leans, captured during a persuasion conversation. It matters because the GOTV universe is built from IDs. A campaign that never IDs voters has no way to build a turnout list and ends up chasing everyone.

How early should a campaign start canvassing?

Early enough that the identified universe is large enough to build a GOTV list from, which for most down-ballot races means months rather than weeks. The common failure is starting so late that the campaign reaches election week with no IDs.

How many volunteers does a field program need?

Work backwards from coverage using the arithmetic above, then inflate for no-shows. There is no useful rule of thumb, because it depends entirely on universe size and how many shifts remain.

Where to go next

The doorstep craft underneath all of this, including training, objections, and safety, is in the complete guide to door-to-door canvassing. The turf-cutting and measurement side is in the canvassing operations guide.

For the software, Beacon runs the field side of a campaign specifically: canvassing software for political campaigns, with voter file import, precinct cutting, walk lists on canvasser phones, and GOTV coverage on a live dashboard.