The win-number worksheet, how many votes does your race really take

Ask a first-time candidate how many votes it takes to win and watch what happens. The eyes go to the ceiling. The mouth says "a lot." Somebody mentions "building momentum." Nobody reaches for a pencil. Folks, a campaign that cannot name its number is a road trip with no destination, and the only thing it will certainly produce is a nice set of yard signs and a bad night.

The number is not a mystery. It is arithmetic you can do at the kitchen table in an hour, using public records that the county election office will hand you for little or nothing. Everything else in a campaign, the doors, the phones, the mail, the money, is a way of reaching that one number. So let's calculate it properly, in order, with every line shown. If you do the worksheet once, you will understand your race better than most people who have been talking about it for years.

Why one number beats a hundred slogans

A campaign has only so many hours, so many volunteers and so many dollars. The win number tells you how to spend all three. If you need 816 votes, you do not need to be loved by 5,000 people. You need 816 of the right ones, and you need to know who they are.

Here is the strongest objection, stated fairly. Some people say that a campaign is about ideas and character, and that reducing it to arithmetic is cynical. It is a decent objection. Ideas matter. But the idea that moves nobody to the polls does not become law. The arithmetic is not an alternative to your message. It is the delivery truck. Without it, the best message in the county sits in the garage.

There are three terms to define before we start:

• Ballots cast means the number of people who actually voted in a given election in your district.
• Votes cast in the race means the ballots that contain a choice for your office. Some voters skip it.
• Win number means the smallest number of votes that guarantees victory in a two-candidate race: half of the votes cast in the race, plus one.

Everything below is a hypothetical for teaching. The ward, the numbers and the candidate are imaginary. Replace each figure with your own from official records, and check your state's rules, because some places use runoffs, plurality wins, or ranked choices, and those change the final steps.

Gathering the record

Step 1: Pick the comparable elections

The most common error in vote math is comparing the wrong elections. Turnout in a presidential year is a different animal from turnout in a midterm, and both are different from an odd-year local election with nothing else on the ballot.

So pick the past elections that are truly comparable to yours:

1. The same office or the same level of office (a council seat, not the governor).
2. The same kind of year (a spring municipal election to a spring municipal election; an odd-year November to an odd-year November).
3. The same district lines, or as close as you can get. If the lines moved, ask the election office for results re-tallied to the current lines, or recompute from the precinct results.

Your sources are the official canvass, the certified results the county or city publishes after each election. Search the county election office website for "official results" or "canvass." If the website is thin, call the office. Most clerks are glad to help a candidate who is doing homework.

Take three to five elections. More is better, but anything older than about ten years starts to reflect a different electorate.

Step 2: Write down the ballots cast

For each comparable election, write the number of ballots cast in your district. Then add a column for the other election types in the same years so you can see the pattern.

Here is the hypothetical. Picture Ward 4 of an imaginary city called Fairmont. Its council seat is on the ballot in odd-numbered years, in November, when nothing else of consequence is being decided. The election office's canvass shows:

The comparable elections for a council race are 2021 and 2023. Those are your two data points.

Step 3: Measure the off-year drop

Now watch what the table says. People who vote in a presidential year do not all vote in an odd year. That drop is the single most important thing a first-time candidate misunderstands, because it tells you that the electorate in your race is a smaller and different group from the one on cable news.

Measure it two ways.

From presidential to the odd year that follows:

• 2020 to 2021: 4,520 minus 1,750 equals 2,770 fewer ballots.
• 2,770 divided by 4,520 is about 0.613. The drop is roughly 61 percent.

From the midterm to the odd year that follows:

• 2022 to 2023: 2,940 minus 1,610 equals 1,330 fewer ballots.
• 1,330 divided by 2,940 is about 0.452. The drop is roughly 45 percent.

The lesson: do not plan on the presidential crowd. In Fairmont, more than half of the people who vote in a presidential November stay home in the odd-year November. Anyone who plans their race around the larger electorate will overestimate the votes in play by a mile, and spend money chasing people who were never going to show up.

Building the number

Step 4: Estimate the ballots you expect

Average the comparable elections:

• 1,750 plus 1,610 equals 3,360.
• 3,360 divided by 2 equals 1,680 expected ballots.

That is a sensible baseline. If turnout has been sliding election to election, the lower number might be the better guess, and you should say so in your plan. Here the decline from 1,750 to 1,610 is 140 ballots, which is small enough to average. If something about your year is different, such as a hot ballot measure, a contested mayor's race, or a school levy, adjust up or down and write down why.

A warning about the temptation to be hopeful: set the baseline on what the record shows, not on what you wish would happen. You can always be pleasantly surprised on election night. You cannot be pleasantly surprised if you spent the last two weekends knocking on the wrong doors.

Step 5: Subtract the skippers

Not everyone who casts a ballot votes in every contest. Some leave the council race blank. Election officials call these undervotes. The canvass often shows them as a line for "undervotes" or lets you compute them by comparing the ballots cast to the total votes counted in that race.

Suppose that in 2021 and 2023, about 3 percent of ballots in Ward 4 had no choice marked in the council race. (Hypothetical. Compute your own from the canvass.)

• 3 percent of 1,680 is 1,680 times 0.03, which equals 50.4. Round to 50.
• Votes cast in the race: 1,680 minus 50 equals 1,630.

Step 6: Find the 50 percent plus one line

In a two-candidate race you win with more than half of the votes cast in the race.

• Half of 1,630 is 815.
• 815 votes is exactly half, which is a tie, and a tie is not a win.
• So the win number is 815 plus 1, which is 816.

The "plus one" is not decoration. The simple rule: take half, drop any fraction, and add one. Check it against an even and an odd total. For 1,630, half is 815, and 815 plus 1 is 816. For 1,631, half is 815.5, drop the fraction to get 815, and 815 plus 1 is 816. Either way, 816 votes beat the other side and 815 does not always.

A note on multi-candidate races. If three or more people are running and your state decides by plurality, the winner needs the most votes, not a majority, so the true number can be well below half. If your state requires a majority and a runoff, you run two campaigns with two win numbers. Look up your state's rule on the election office or legislature website, and do not guess.

Step 7: Add a margin of safety

Nobody wins by exactly one vote on purpose. Polls, records and plans all miss. Absentee ballots run late. Some of your commitments will forget. You want a cushion.

A cushion of 5 percent of the votes cast in the race is a reasonable planning figure for a first campaign. It is a choice, not a law. Use more if your race is volatile.

• 5 percent of 1,630 is 81.5. Round up to 82.
• Target: 816 plus 82 equals 898 votes.

Write 898 on an index card. That is the number you are working to. Everything is measured against it.

Turning the number into work

Step 8: Count the votes you already have

You are not starting at zero. Some voters will support a candidate of your party no matter what, and some know you personally. Estimate your floor from the past:

• In 2021, the candidate on your side got 690 votes in Ward 4.
• In 2023, the candidate on your side got 640.
• Use the lower number, 640, because turnout was lower and you are planning on lower turnout.

This is a rough floor, not a promise. Candidates differ. Still, if the last two candidates on your side drew 690 and 640, a reasonable plan starts near 640.

The gap: 898 minus 640 equals 258 votes you must add.

Step 9: Split the gap into two jobs

Every vote you add comes from one of two places:

1. Turnout. People who agree with you but do not reliably vote in an odd-year election. You do not need to change their minds. You need to get them to the polls.
2. Persuasion. Voters who do vote, and who have not made up their minds or lean the other way. You need them to choose you.

Divide the 258 between them based on what your voter file shows. A reasonable first guess for a Republican candidate in a ward that leans the other direction: lean on persuasion more. For a ward that leans your way, lean on turnout. Say, hypothetically, the file shows plenty of likely supporters who skipped the last odd-year election:

• Turnout target: 110 votes.
• Persuasion target: 258 minus 110 equals 148 votes.

Check: 110 plus 148 equals 258. Good.

Step 10: Turn the targets into contacts

Now the part where the worksheet pays for itself. You know the number of votes. How many conversations does it take to get them?

You do not know this yet, so use a planning assumption and replace it with your own results after your first week of work. Say you assume, hypothetically:

• One in two contacted supporters who usually skip odd-year elections will actually vote after you reach them.
• One in four contacted persuadable voters will end up voting for you.

Then:

• Turnout: 110 votes at one vote per two contacts is 110 times 2, which equals 220 contacts.
• Persuasion: 148 votes at one vote per four contacts is 148 times 4, which equals 592 contacts.
• Total: 220 plus 592 equals 812 contacts.

Over a six-week campaign, 812 divided by 6 is about 135.3, so roughly 136 contacts a week. If you knock doors and one door in three yields a real conversation, the doors you must attempt are 812 times 3, which equals 2,436. That sounds large until you divide it among a dozen volunteers over six weeks. Twelve volunteers over six weeks is 72 volunteer-weeks, and 2,436 divided by 72 is about 34 doors per volunteer per week.

Suddenly, "win the ward" is a schedule. A schedule can be managed. A mood cannot.

Take that to the next steps. Your cost for reaching those contacts is a different worksheet, and a campaign budget worksheet for the candidate with no money and no staff picks up exactly where this one ends. The method for the doors is in door-knocking that works, a walkthrough from the turf map to the thank-you, and the closing stretch is covered in the last seventy-two hours, a get-out-the-vote checklist.

Run it twice if there is a primary

Many seats are decided in the primary. In a one-party district, the primary may be the real election. In a party primary, the turnout is a fraction of the general, and the win number is smaller. Same worksheet, different inputs: pull the past primary canvass, find the ballots cast in your party's primary, subtract undervotes, and take half plus one.

If, hypothetically, 520 voters cast a Republican primary ballot in the ward in the last comparable year and 4 percent skipped your race:

• 4 percent of 520 is 20.8, round to 21.
• Votes in the race: 520 minus 21 equals 499.
• Half of 499 is 249.5, so the win number is 250.

That number is something a determined candidate and a handful of volunteers can reach with a few hundred doors and a good phone list. This is why Why every Republican owes the primary election a vote, every time is not a sermon but a piece of arithmetic. If the people you are trying to reach do not vote in the primary, the general election you want may never arrive.

Also check whether your state limits primary participation by party registration. Can I vote in the Republican primary? Your registration questions answered explains why the rules differ, and your voter file will show who is eligible to be reached.

The mistakes the worksheet catches

Most losing campaigns can trace some of their trouble to a wrong number. A few worth naming:

• Using the presidential electorate. Shown in Step 3. It inflates the win number by a lot.
• Forgetting the undervote. Small, but at the margin it matters.
• Confusing registered voters with likely voters. The voter file may list 6,000 people. The record shows 1,680 usually vote. Plan on the second number.
• Counting friends and family as the base. Your cousins are lovely. They are probably not in the ward.
• Never updating. After your first two weeks, replace every assumption in Step 10 with your actual log of contacts and results. The worksheet is a living document.

Keep the worksheet in a folder with the date. When the next candidate in your county asks how you did it, hand them a page and a pencil. A party that teaches this to its candidates wins more races than a party that tells them to believe in themselves.

The bottom line

• Calculate your win number before you spend a dollar or print a sign: ballots expected, minus undervotes, half plus one, plus a margin of safety.
• Use comparable elections, the official canvass, and the true off-year drop, not the presidential crowd.
• Subtract the votes you already have, split the gap between turnout and persuasion, and convert it to contacts per week.
• Replace every assumption with your own results after two weeks, and keep a dated copy of the page.
• If your seat is decided in the primary, run the worksheet again for the primary, and check your state's rules for runoffs and plurality wins.