State of the Field

This paper describes what the GospelGraph data holds about the United States as a whole, rather than about any one neighborhood. It covers 83,344 census tracts, 98.7% of the national index, and 125.7 million households. Eight findings stand out, and they are not all in the same direction.

Shalom v2 scores five domains into a composite: provision and work, justice and protection, life and health, household and generations, social fabric. Need pressure is its inverse: 100 minus Shalom. It is the level under the movement.

Field Guide v4.5 is the household layer: thirty core guides describing who lives in a tract and in what mix, plus fourteen context modifiers that describe a condition sitting on top of the household type rather than replacing it. Every tract carries a composition rather than a label, and this paper weights by that composition wherever household types are compared.

The first thing the distribution says is that the familiar framing is wrong. Gospel access is not a country split between well-churched suburbs and abandoned cores. It is one broad hump with very little at either end. Only 1,321 tracts in the United States reach the best tier, and only 837 fall into the worst. Between them, 81,186 tracts, 97.4% of the country, sit somewhere in the middle three.

Read the last two columns together. The average number of congregations inside a tract is essentially flat from Watered to Arid, at 7.4, 6.8, 6.8 and 7.3, and only falls at Desert. Median distance to the nearest congregation, meanwhile, rises ninefold across the same span, from a seventh of a mile to over one and a quarter.

Access is a question of how far apart the churches and the people are, which is a question about how the ground was built.

8,899 tracts have no congregation inside them at all, which is 10.7% of the country, holding 11.6 million households. Another 7,938 have nothing within two miles.

The state table contains a distinction the single score hides. Utah, Nevada, Arizona and Idaho are dry and empty: between a fifth and two fifths of their tracts contain no congregation at all. Mississippi, Tennessee, Arkansas and Kentucky score almost as dry with virtually every tract churched, at 2 to 5% empty.

These are not the same problem. The Mountain West needs presence where there is none. The upper South has presence everywhere and access that is still poor, which means the constraint is something other than the number of buildings. Any national strategy that treats a GDI of 54 as one condition will send the wrong worker to one of the two.

Across the country, where a direction can be established at all, improvement outnumbers decline by better than two to one: 38,861 tracts confirmed opening against 18,287 confirmed hardening.

The sector's default posture is decline, and the data does not support that as a general claim. It does carry the qualification that follows.

The cross-tabulation is the most important table in this paper. The share of tracts improving falls at every step as access worsens, from 50.7% to 41.7%, and the share declining rises at every step, from 19.6% to 25.1%. There is no reversal anywhere in the sequence.

So the encouraging national ratio is unevenly distributed. The places with the best gospel access are also the places most likely to be improving on their own, and the places with the worst are the least likely. Whatever is producing improvement across the country is doing least where it is needed most.

The gradient is not enormous, nine points of spread from best to worst, and it should not be over-read. But it runs the same way at every step, which is what makes it worth trusting.

Need pressure across the country has a median of 50.3, with a quarter of tracts below 39.9 and a quarter above 60.9. The tenth and ninetieth percentiles are 31.9 and 69.3. 7,568 tracts sit at 70 or above, which is 9.1% of the country and 9.5 million households. That is the range where several domains are usually failing at once.

Crossing need against viability produces four situations, and they are more evenly balanced than one might expect.

The quadrant worth naming is high need with low viability: 17,545 tracts, one in five. Conditions are hard and there is no third place, no walkable intersection, no public room. A conventional plant, which rents a space, advertises a service and gathers a crowd, has the least to work with here. These tracts are almost never distinguished from ordinary hard ground in the way work is assigned.

Its opposite, high need with high viability, is nearly thirty percent of the country and is the largest single quadrant. Hard conditions, but somewhere to gather. If there is a place where conventional planting is most likely to succeed against real need, that is where it is.

Note also that three quarters of American tracts are suburban or sparse-urban. Dense urban ground, the setting most church-planting literature assumes when it talks about cities, is 8.0% of the country.

The household layer produces the most lopsided distribution in the paper. 50.6% of American households fall in LifeMode group L1, and the two largest guides, Affluent Owner Family Suburbs and Exurban Working Homeowners, account for 27.3% between them. The four largest cover 46.9%.

This has a blunt implication. A ministry strategy that imagines itself serving a wide variety of household types is, in practice, mostly serving suburban homeowners, whether or not it says so. And a resource built for the tail will rarely be the one that gets used. The L3 group is 0.7% of households.

The largest modifier is public housing, present in 12,804 tracts and covering 9.9 million households. That is close to one American tract in seven, and it is worth sitting with, because public housing is usually discussed as an exception rather than as a seventh of the country.

Displacement pressure and health burden between them touch nearly 30 million households. Military adjacency, at 11.2% of tracts, is larger than most denominations' entire strategic attention to it.

The recruiting language points away from the need

Church planting recruits with urban imagery: the storefront, the walk-up, the transit stop, the neighborhood with a main street. That is a real need and it is eight percent of American census tracts. Three quarters of the country is suburban or sparse-urban ground, and the fastest-opening deserts in this dataset are exurban subdivisions and university-town build-outs.

The effect on a young person with a genuine call and no fixed geography is predictable. They hear that the frontier is downtown, they go, and they arrive in the most churched ground in the sample. Meanwhile a subdivision forty minutes out, holding eleven thousand households and one congregation, has nobody making its case because nobody has written the appeal. Somebody has to be willing to say from a platform that some of the hardest and most open ground in America looks like a cul-de-sac.

The corollary is about who is sent, not only where. Pioneer work in a place with no institutions, no memory and no crowd is a different vocation from revitalizing a congregation that has been declining for thirty years. Both are needed at scale, 2,168 tracts for the first and 3,875 for the second, and the two are treated as one job description.

The funding clock does not fit the ground it funds

Support is typically built on a three-year runway with attendance milestones from year one. On the growth edge that is close to the worst available arrangement. A neighborhood with no institutions has no existing crowd to draw on, so the first two years yield relationships and very little attendance. That is the pattern a metrics-driven funder reads as failure, in the places this data says are opening fastest.

What the household guides prescribe for these places is unfundable under most current arrangements, and is nonetheless correct. Hold one gathering on one date, unchanged, for three years, because a place with no institutions cannot distinguish a church from a pop-up until something has survived long enough to become predictable. Stand at the school pickup line weekly, because it is the only moment the neighborhood is gathered and out of its cars. Find the households who arrived in phase one and give them real information to pass on, because in a place with no memory the earliest residents are the nearest thing to elders it has.

Milestones for years one and two should be countable in a different currency: households known by name, consecutive weeks the gathering held its date, phase-one residents who will carry accurate information. None of those is soft, and all of them are countable. What none of them is, is attendance.

Ninety minutes, and what the income figure hides

The household guide for the type most over-represented among the unreached warns against a specific misreading: a median income near a hundred thousand dollars, on the growth edge, is servicing a new mortgage, two cars, childcare and a fifty-minute commute. Upper-middle income here does not mean slack. Two commutes and small children compress the usable evening to roughly ninety minutes.

A church that schedules midweek at seven o'clock in a place like this is not competing with indifference. It is competing with the drive home, and it will lose, and it will conclude that the neighborhood is hard to reach. The guides' instruction is mundane and load-bearing: move everything to half past eight in the evening, or to Saturday morning.

A related warning is worth repeating because it is so easy to get wrong with the best intentions. Do not admire the commute. Calling a man's fifty-minute drive heroic makes it harder for him to say that it is grinding him down.

A fifth of the country cannot host a conventional plant

17,545 tracts, 21.1%, combine high need with low missional viability. Hard conditions, and no third place, no walkable intersection, no public room. The standard approach assumes an infrastructure that is not there.

These tracts are almost never separated out when work is assigned. They receive the same playbook as a tract with a main street, the playbook fails, and the failure is attributed to the planter rather than to the assignment. That is a category error with a human cost, and it is visible in advance for anyone who looks.

What fits instead is not a secret: work that travels rather than gathers, that meets in houses rather than rooms, that accepts a longer horizon and smaller numbers. What is missing is permission to plan that way from the beginning rather than arriving at it after two years of trying to rent a storefront that does not exist.

Affluence mistaken for reach

High-Achievement Homeowners are scarce among tracts unreached beside a church (0.54×) and over-represented among tracts with no church at all (1.62×). The natural reading of the first figure, that these households are covered, is wrong. It is wrong in a direction that costs nothing to believe, which is why it persists.

The guide for these households locates the mechanism somewhere other than programming. It warns against taking the money first, because it is the fastest way to fund anything and it settles what those households are understood to be for, which is what they were already assuming. It warns against thanking them from the front, because public gratitude confirms the arrangement and makes it harder for anyone to approach them in any other role. It warns that the sermon against materialism is the expected one, costs the preacher nothing, and lands as noise.

Its prescribed action is a direct rebuke to standard practice: decline one offer of funding this quarter and ask for a year of somebody's Tuesdays instead. Give a young person here a job that can fail, and let it. Say something unfinished about your own life at a table where nobody does that.

A congregation that has only ever asked its wealthy households for money has not reached them. It has billed them, and both parties have found the arrangement comfortable.

What to say to a pastor on the hardening list

3,160 congregations sit inside a tract that is confirmed to be getting worse. Handed carelessly, that figure says: you were here and it declined. That is not what it says, and anyone who presents it that way has misunderstood the measurement and the ministry alike.

What it says is that faithfulness and outcome came apart. The tradition has always insisted those are different things, right up until somebody produces a chart. A congregation is not a labour market or a housing policy, and most of what moves a tract's conditions over a decade is outside any church's reach.

The honest use of this list is company. Three thousand congregations are in the same position, which means no pastor in one of them is the anomaly they believe themselves to be. That is the first thing to say, and it should be said before any number is.

The second is a question about posture rather than effort. A church that has been in a declining neighborhood for a generation organizes itself around holding on: keep the building, keep the schedule, keep the remaining members. Each instinct is reasonable and together they produce a congregation that is present but no longer reachable, which is the gap between presence and access this measurement exposes. The Sunday service survives and the front door narrows.

So the question is whether anything we do is still addressed to the people who now live here. In a tract that has hardened over ten years, the households outside are frequently not the households the congregation was built for. Nobody chose that. It happened at the speed of a housing market, which is faster than a church changes.

Nobody prays for a tract number

94.4% of tracts now carry a curated name. This looks like presentation and is not. Every act of ministry this data exists to serve happens under a name: what is prayed for, preached about, printed in a bulletin, said by a person asked where they are from.

The growth edge sharpens the point. A neighborhood named for a century carries meaning and every accumulated offence: Eastport, Southmost, Warrendale. A subdivision carries a name a developer chose to sell houses, arriving before any of the shared life that usually produces one. Both kinds sit in the tables above. Turning the second sort of name into the first is a fair description of what planting on the growth edge consists of.

A pastor handed a tract score, a movement tier and a viability index can conclude that their neighborhood has been graded and that they have been graded along with it. That reading is available in every figure in this paper and it is false in all of them.

None of this measures faith. None of it measures whether a congregation is good, whether a pastor is faithful, or whether God is at work on a particular street. It measures how far people are from a church, which direction the conditions around them are moving, and whether there is a room. Those things are worth knowing. They are not the thing anyone cares about, and a paper like this one is only useful to the degree that it keeps that distinction in front of the reader.

Numbers can do one narrow thing. They can stop a denomination looking only where it already was, and they can tell a tired pastor that three thousand others are standing in the same place. That is the whole of it, and it is not nothing.

Two choices in the method shape every count above, and both belong before the caveats.

Movement is read across both confirmed tiers. The long-baseline model reports opening and hardening each at two speeds, and the fast tiers are not small: opening_fast holds 7,666 tracts and hardening_fast holds 3,577. A query that reads only the base tier stops at roughly +6.8 points per four years and loses the steepest movement in the country.

Special-context tracts are counted. 18,066 tracts, 21.7% of the country, carry a flag naming them as campuses, senior communities, or mixed and institutional populations. The map prints a caveat beside those tracts and then shows them, and this paper does the same. People live in all of them. Excluding them would also drop three of the fourteen context modifiers to almost nothing, since farmworker concentration, senior living and university town live largely inside those tracts.

What is not measured

None of this measures faith, conversion, discipleship or the health of any congregation. It measures access, meaning whether a church is near enough and stable enough to be reachable, and it measures the movement of the conditions people live under. A tract can be full of believers and score badly.

Vintage lag

The five-year American Community Survey trails reality by roughly three years at the midpoint. A subdivision finished in 2023 is thinly represented. Where this paper says a place is new, it is likely to be newer than the figures show.

Church counts

Congregation presence comes from a compiled index rather than a census of churches. It will miss house churches, campus ministries and congregations meeting in rented rooms, all of which bias access downward in the places most likely to have them.

Circularity

The access score counts church presence, so any statement of the form “a church is here and access is still poor” is partly self-referential. Comparisons between household guides survive it, because the circularity is identical for all thirty. Absolute rates should be treated with more suspicion.

Thresholds are ours

Tier boundaries, the desert floor of 60, the unreached cut of 55 and the viability midpoint of 50 are choices. Moving them moves every count in this paper. None of the directions or gradients depend on them.

What would sharpen it

Church-plant announcements, to compare where planting goes with where the need is. A closures database crossed with trajectory, to learn what the years before a closure look like. Both are event data; almost everything here is state data, and events are what give direction.