Twenty thousand people is not a number of doors

The claim in the wild

Nearly every strategic plan a congregation writes contains a sentence like this one:

We serve a neighborhood of about twenty thousand people.

Nobody says it carelessly. It comes from a real place — a Census pull, a city website, a consultant's slide, a memory of what the neighborhood used to be. It is usually accurate as a count of people. It is almost always useless as a description of a mission field, because a congregation does not reach people. It reaches households: a door, a table, a family, a person living alone with a name you learn.

So this week's question is a boring one with an expensive answer. Twenty thousand people is how many doors?

What we checked

The count of residents and the count of households, ZIP code by ZIP code, from the Census Bureau's American Community Survey 2024 five-year estimates. Two tables: population (B01001) and household type (B11001), joined on the same geography, plus a third (B11005) for households with children.

To keep the comparison honest, the figures below use one universe throughout: 21,393 ZIP codes with at least 1,000 residents and 500 households. That threshold sets aside institutional places — a campus, a prison, a care facility — where the relationship between people and homes breaks down. Every number printed here uses that same denominator, so the numbers in this piece add up against each other.

What the numbers say

Across those 21,393 ZIP codes, the typical household holds 2.52 people. That median hides a real spread: at the 10th percentile a ZIP averages 2.15 people per household, and at the 90th it averages 3.03.

Now hold the population constant. There are 112 ZIP codes in this release holding between 19,803 and 20,200 residents — very nearly the same twenty thousand people in each. Their median household count is 7,826.

But the 10th percentile of that group has 6,665 households, and the 90th has 9,374. Identical population: 41 percent more front doors at one end than at the other.

Two real ZIP codes make the point better than a percentile does.

46202 — downtown Indianapolis. 20,180 residents, 10,786 households, 1.87 people per household.

33974 — Lehigh Acres, Florida. 19,927 residents, 5,438 households, 3.66 people per household.

The populations sit within a few hundred residents of each other. The number of households does not. A church in 46202 and a church in 33974 are handed the same population figure and two completely different shapes of work.

Why the doors look different

The same two ZIP codes, described by who is behind the door:

• In 46202, 54.8 percent of households hold one person. 9.2 percent hold children.
• In 33974, 12.1 percent hold one person. 50.1 percent hold children.

Nationally, across the same 21,393 ZIP codes, the median share of single-person households is 27.5 percent — more than one in four doors. The 10th percentile is 16.9 percent; the 90th is 39.2.

Households with children follow the same pattern in reverse. Median 28.6 percent of households, with a tenth of ZIP codes below 18.4 percent and a tenth above 39.5.

Two program decisions hide in those three sentences, and neither is a matter of taste. A congregation whose neighbor is 46202 and whose calendar is built around family ministry is building for fewer than one door in ten. A congregation in 33974 that reads "young adults don't join churches anymore" and concludes the neighborhood has no families is describing a different neighborhood than the one it is standing in.

What this doesn't tell you

Four limits, stated before somebody else states them for us.

A person in a dorm is not a household. The Census calls all the people occupying a housing unit a household; people living in group quarters — dorms, barracks, nursing facilities, correctional facilities — are counted as population but not as households. That is why this release's outer edges look the way they do: one ZIP averages 1.19 people per household and another averages 11.11, and both are almost certainly institutional populations rather than neighborhoods. We do not yet ingest the group-quarters table (B26001), which is the missing piece that would let us set those aside and publish a clean range instead of a caveated one. That gap is on the list. Until it closes, read the quartiles and ignore the ends — the number of households in a 20,000-resident ZIP still ranges from 5,438 to 11,448, and both of those are real places.

A ZIP code is not a neighborhood. It is a postal and statistical boundary drawn for mail. It can split one walkable district across three ZIPs and merge a river away from its own congregation. These figures describe the shape of a place, not the reach of a particular church.

Five years of people, averaged. The ACS pools 2020–2024 into one estimate. It is the right instrument for a neighborhood and the wrong one for a street corner, a single building, or a change that happened last spring. Every estimate also carries a margin of error; a ZIP with a few thousand residents can have one wide enough to swallow a real difference.

Percent of households is not percent of people. A congregation counting heads will find different proportions than a congregation counting doors. Both are true. They answer different questions, which is the entire point of this issue.

What to do with it

Take the population figure out of your plan and replace it with two: the number of households, and the number with children under eighteen. If you do not have those, you do not know your neighborhood — you know its headcount.

Then ask the question the headcount cannot answer. If more than a quarter of the doors near you hold one person, what in your building was designed for them — not for a family that might attend, but for the person who already lives there? If fewer than one door in ten holds children, as in a ZIP that looks like 46202, then ministry with children is not a program that failed; it is a program aimed at a smaller and more precise target than anyone told you.

The data does not make that decision. It only refuses to let the decision be made on a number that was never about doors.

Sources

• U.S. Census Bureau, American Community Survey 2024 five-year estimates (2020–2024), tables B01001, B11001 and B11005. Read through the Sanctuary warehouse; every figure above has a named query in the receipts table below and can be re-run.
• Census Bureau, ACS subject definitions and guidance — for the household/group-quarters distinction used in "What this doesn't tell you."
• Table B11001 itself, for anyone who wants to read it without us: data.census.gov.

Nothing in this issue relies on an unverifiable claim. Where we cite a third party we link it and we say so; see what we compute and what we only cite.