Seven church health metrics carry almost any board meeting: attendance frequency by household, first-time guest return, giving participation, giving concentration, volunteer participation, group participation, and attendance year over year. Each is a rate rather than a total, and each has names behind it.
Key points
- A board governs rather than operates: the monthly pack is seven rates and their direction, not a page of totals, and every rate should decompose to a list of households.
- Groups is the only Planning Center product that calculates a participation rate for you: the Reports page shows a church participation percentage beside average group size and turnover. Planning Center doesn’t document which population it divides by, though, so the board figure still comes from an export against your own adult-attender count.
- Giving’s Total Donated dashboard widget includes donations that are uncommitted or not fully processed, so a board figure read straight off it can move for bookkeeping reasons alone.
- Board members given Giving’s Reviewer permission see the dashboard but not reports, donation history or payout reports, so the person asking the question can’t pull the answer.
- Two of the seven have a defensible national figure: 44% of participants volunteer and 44% of attenders are in a group. Giving concentration has no current benchmark, only a 1990s literature to check the shape against.
The board pack I see most often is one number long: average weekly attendance, this month against last, a giving total, a note about the roof. Nobody chose that. It’s what falls out of the tools. The number is easy to produce and everything else takes an afternoon.
Then somebody asks whether the church is growing and the room goes quiet, because a total can’t answer that. Here are seven I’d put in front of a board instead.
What makes a church health metric worth a board’s time?
Three tests, and most of what ends up in a board pack fails at least one.
It has to be a rate, not a total. Totals cancel. Four hundred came last month and four hundred this month, and in between eleven families left and eleven arrived, and the arithmetic hides both. A rate has a denominator, which forces you to say what you’re measuring against, and that’s where the honesty in a metric lives.
When it moves, you have to be able to name who moved it. If giving participation drops two points, the useful next step is a list of the households that stopped; if that list can’t be produced, the metric is decoration. It’s the test the church metrics dashboard is built around, and it eliminates most vanity numbers.
It has to be reproducible in under an hour. A metric that rests on one staff member’s spreadsheet skills lasts as long as that staff member.
One constraint before you start. Board members with Planning Center logins have usually been given Giving’s Reviewer role, and the permissions table is specific: Reviewer sees the dashboard, not payout reports, the donation list or a donor’s history. So whoever assembles the pack has to bring the decomposition with them.
A fuller list is in church metrics that matter. This is the subset a board should see.
1. Attendance frequency by household
What it measures. Not how many came, but how often each family comes. Households disengage by attending less long before they stop, which makes this the earliest signal on the list.
Formula. Of the households that were already active when the window opened, the share that attended at least seven of the last thirteen Sundays, beside the previous quarter’s figure. Households whose first check-in falls inside the window can’t reach seven however faithfully they come, so a good month for newcomers would read as falling frequency; hold them out until they have a full window behind them.
Where it comes from in Planning Center. Check-Ins records every check-in individually, so the raw material is there. The report isn’t: the event chart exports counts by date, while the people come from a custom report on the event’s Reports tab. Nothing rolls check-ins up to a household and nothing compares one household across two windows, which is the whole of Planning Center attendance reports.
What healthy looks like. No national benchmark exists for frequency, and I’d distrust one if it did exist, because it turns on how a church defines an active household. What national data does say is that regular no longer means weekly: Pew’s Religious Landscape Study found one-third of US adults say they attend in person at least monthly, including 25% at least weekly.
When it moves. Falling frequency with a flat total means departures masked by arrivals. Ask for the households whose frequency dropped, sorted by size of drop, and give the list to whoever knows them.
2. First-time guest return rate
What it measures. How many first-time guests come back. A guest count measures one Sunday’s welcome; the return rate measures whether anything happened afterwards.
Formula. Of the people whose first visit fell in a given month, the share who attended again within 28 days of their own first visit. Report it on a one-month lag, so every guest’s window has closed.
Where it comes from in Planning Center. The Check-Ins First time visitors widget shows people on their first check-in for an event, or their first check-in ever, and excludes one-time visitors. A custom report scoped to first timers only gives the names. The return half is a People list with two conditions: a first check-in in the month you’re measuring, and a second check-in between the start of that month and four weeks after its end, run once on that closing date. Not “any check-in since”: a list can’t express a window relative to each guest’s own first visit, and an open-ended condition counts returns from months later. The bounded version, and the export route for an exact per-guest rate, are worked through in first-time guest tracking. One trap decides whether any of it works: guests checked in through label-only mode are never saved to People. First-time guest tracking covers capture.
What healthy looks like. Nobody credible publishes this. Lifeway Research measures the mechanism instead, reporting that 80% of Protestant churches use printed guest cards and 38% an online form. The sector counts how it collects the card, not whether the guest came back. Every “X% of guests return” figure in circulation traces to a vendor blog.
When it moves. Faster than anything else here, and in response to staffing and process. If it falls, the question is what changed about the first ninety seconds and the first week, not the sermon.
3. Giving participation rate
What it measures. The share of your church that gives at all. Boards are shown totals and almost never participation, which is the number that says whether generosity is broad or propped up.
Formula. Active households with at least one gift in the quarter divided by active households. Match donor keys against the active list, not every donor in the export, or the rate has no ceiling. Quarterly, not monthly. A month is too short for irregular givers.
Where it comes from in Planning Center. The By donor tab on the Donations page rolls a date range up per donor and exports to CSV like every other Giving report. The Home dashboard’s Donors widget shows the number of donors who gave during a timeframe, which counts every donor rather than the ones on your active list, so it is a sanity check on the export and not the numerator itself. Giving reports on donors rather than households, so a couple with two cards counts twice unless you join them. That’s one reason Planning Center Giving reports stop short of a board figure.
What healthy looks like. Your own baseline, but know the direction of travel. Hartford Institute’s 2024 congregational finance study reports that clergy in its interviews describe total donor numbers falling while per-person amounts rise, alongside median per-capita income rising from $2,000 in 2020 to $2,222 in 2023. That figure is calculated on total congregational income rather than donations alone, and the report notes it is a real-terms fall, since holding 2020 purchasing power would have required about $2,355. Treat the donor pattern as what practitioners report, not a measured national trend; a steady total on a narrowing base is the case worth checking for in your own numbers.
When it moves. Participation falling while the total holds is the important case, and it always means the same thing: concentration is rising. It is also the point at which a lapsed-giver list earns its keep.
4. Giving concentration
What it measures. How much of your budget rests on how few households. The most useful risk figure a board has, and always higher than the room assumed.
Formula. The share of total giving from your top ten households, and separately from your top ten percent, on a rolling twelve months.
Where it comes from in Planning Center. Export the By donor report for the trailing twelve months, sort by donation total descending, and take a running total down the list. That column is sortable in the report itself. Nothing in Giving computes it; it’s two minutes in a spreadsheet. Join households by hand first. Concentration measured on individual donors understates itself.
What healthy looks like. There is no current benchmark, and I did look. The National Congregations Study doesn’t ask the question, and neither Lake Institute’s national study of congregational finances nor Hartford’s finance research reports it. What exists is older: a University of Notre Dame review of the religious giving literature reports congregational studies from the 1990s consistently finding about 80% of funds from 20% of participants. That’s a shape to check yourself against, as church giving analytics goes through, not a target from a live measurement. What is established is the exposure: congregations received 81% of their income from regular individual contributions as of NSCEP’s 2017 data, and for a third of them those contributions are the only income source.
When it moves. Up needs a plan, and three or four points in a year is a trend wherever it started. If your top ten households carry much of the budget, the board’s exposure isn’t a giving problem, it’s a continuity problem. Church giving analytics covers the fuller set.
5. Volunteer participation
What it measures. The share of adult attenders who serve, and the number of established volunteers whose frequency has dropped. Serving is the first thing to go when someone disengages.
Formula. Adult attenders who also served at least once in the quarter, matched on Person ID, divided by adult attenders. Beside it, a count of volunteers who served three or more times last quarter and once or fewer this quarter.
Where it comes from in Planning Center. Services records whether scheduled people showed up, and team leaders can take attendance only for the teams they lead, so coverage depends on your leaders. The Check-Ins and Services integration makes a scheduled volunteer’s check-in count in both products, which is what makes the numerator trustworthy. The frequency comparison is an export and a pivot, in how to measure volunteer engagement.
What healthy looks like. One of two metrics here with a national figure. Faith Communities Today found nearly half (44%) of congregational participants volunteer in some fashion, and Hartford reported in 2026 that volunteer engagement has returned to pre-pandemic levels. Read 44% as a marker, not a target: the ratio turns on whether you count the twice-a-year greeter like the weekly sound engineer.
When it moves. Down is an early warning rather than a staffing problem. Volunteers rarely quit; they stop being available, and in the churches I’ve worked with serving thins out before attendance does, so treat a sustained fall here as an early warning rather than a staffing problem.
6. Group participation rate
What it measures. The share of your adults in a small group. Someone in a group is known by eight people who will notice if they disappear; someone attending a year with no group can leave without anyone having a reason to call.
Formula. Adults on an active group roster who are also adult attenders, divided by adult attenders. Planning Center calculates a version of it on an undocumented denominator; the board number comes from your exports.
Where it comes from in Planning Center. The one metric with a version you can read off a screen. On the Groups Reports page, the Overview tab’s Health Stats show the total number of groups, average group size, unique members, church participation percentage, and turnover rate for whatever filter you set. One caveat before it reaches a board pack: this percentage is built from membership, not attendance. A group whose leader stopped meeting still contributes every name on its roster, so a stale roster makes participation look healthier than the rooms are. The attendance side has the opposite blind spot: canceled events and events with no submitted attendance are left out of each member’s attendance percentage. Neither number alone tells you a group has quietly stopped. Small group metrics goes through the rest.
What healthy looks like. Lifeway Research’s survey of groups ministry leaders found that around 2 in 5 worship attendees at the average church (44%) also typically participate in small groups. That’s the closest thing to a usable benchmark on this list, and it still depends on what your church counts as a group.
When it moves. Slowly, in both directions. A single month is noise; a two-quarter slide is real. Falling while attendance holds means you’re gaining attenders without connecting them.
7. Attendance year over year, not month to month
What it measures. Direction, with the seasonality taken out. It answers “are we growing”, and it’s the one most often replaced by a month-on-month comparison that answers nothing.
Formula. This month against the same month last year, plus a rolling thirteen-week average against the same thirteen weeks a year earlier.
Where it comes from in Planning Center. Headcount reports on an event cover up to 53 past sessions. That’s a year of Sundays plus one, so a single report holds this week against the same week last year, and nothing more. The rolling 13-week average against the same thirteen weeks a year earlier needs about 65 weekly sessions, so that half of the metric means exporting counts by date and keeping the file; church attendance trends sets the export at 24 months for exactly this reason. A person’s Activity tab covers the past twelve months, one person at a time.
What healthy looks like. Flat is better than it sounds. Median weekly worship attendance across US congregations fell from 137 in 2000 to 65 in 2020, and Hartford’s EPIC project reported in April 2026 that it had risen for the first time in 25 years, to 70, while cautioning that nearly half of congregations are still declining.
When it moves. Two consecutive quarters down year over year is a board conversation. One quarter is weather, and church attendance trends covers reading the line, while how to track church attendance covers getting a number worth trending.
How do you build the monthly board pack from Planning Center?
The honest version, tedious parts included. Half a day the first month, an hour after that.
- Fix the windows and write them down. They differ by metric, and the formulas above already say which: thirteen Sundays for attendance and retention, a calendar month for the guest cohort, a quarter for giving participation and serving, a trailing twelve months for concentration, and this-month-against-last-year plus a rolling thirteen weeks for attendance year over year. Whatever you pick, keep it. A series calculated three ways is not a series.
- Define “active household” once and print the definition in the pack, so nobody re-litigates it at the meeting. Any check-in or gift in the last twelve months is defensible. For the frequency metric, membership in the cohort is decided at the window’s start: a household is in if it had qualifying activity, a check-in or a gift, in the twelve months ending the day the thirteen weeks began. Not at any time ever: a family with one check-in in 2023 that reappears this quarter is a returner, not a regular, and can’t reach seven Sundays either. One that first appeared inside the window, whether by check-in or by a first gift, can’t reach seven Sundays yet, and leaving it in turns a good month for newcomers into a falling rate.
- Export person-level check-ins, not chart data. And the chart data too, for the year-over-year line. Build a custom report as CSV from the event’s Reports tab for everything household-level; the event chart’s CSV is counts by date and won’t pivot. For metric 7 export that chart separately, timeframe set to 24 months, so the rolling thirteen weeks has the same thirteen a year earlier to sit against.
- Match on Person ID, never on name. Two Sarah Johnsons and a Mike who is Michael in half the rows will corrupt everything downstream. Household ID doesn’t come through the People export, so collapse households by surname and address by hand.
- Reduce to household-and-Sunday pairs, then pivot onto the active-household list, not onto itself. Export the active households from step 2 as their own file first. Assign the household key on the raw check-in rows and collapse them so a household counts once for any Sunday any member attended, before pivoting; then look the pivoted counts up from the active list, keeping zero for every active household with no check-in in the window. Pivot the check-in rows alone and the denominator silently becomes households that attended, which is a different and flattering number. Pivot per person first and the dates are gone, so you can’t tell a couple who alternate Sundays from one who come together. Then count distinct Sundays per household, never check-ins, or a family with three kids counts three times.
- Export By donor once per window. The trailing twelve months for concentration, and the quarter, this year and last, for participation. For concentration, aggregate donor rows to one per household first, using the household key from step 4, then sort descending and add a running total column. For participation, match each quarter against an active-household list as of that quarter, not today’s, or last year’s rate is divided by households that weren’t there yet and misses the ones that have since left. Sorting the raw export ranks donors, not households, and a couple with two records splits one share into two.
- Compute group participation from your own files, and read the Groups Reports Overview beside it. Planning Center doesn’t document which population the on-screen percentage divides by, so don’t assume it matches the adult-attender denominator the rest of the pack uses. It isn’t the number in the formula above. Export active group members, intersect their Person IDs with the step 3 attenders, and divide by the adult attenders. The small group metrics how-to walks through it. Then pull serving from Services, checking the Check-Ins integration is on first.
- Keep the file, not just the slide. Month one is worth nearly nothing. Month fourteen is worth something, because you can compare it to month two.
Step 8 is where this dies. Not because it’s hard, but because it happens again next month, and the person who built the spreadsheet has the least free time. If you’re keeping it alive by hand, exports into Google Sheets on a schedule is the version that survives longest.
The one-page board report
Everything above on a page: two columns of numbers, this period and the same period last year, one line of interpretation each.
| Metric | How it’s computed | Where the data lives | Cadence | Benchmark |
|---|---|---|---|---|
| Attendance frequency by household | Households attending 7+ of the last 13 Sundays ÷ households already active when the window opened | Check-Ins custom report, pivoted | Quarterly | None; your own baseline |
| First-time guest return rate | First-timers in month M who returned within 28 days of their own first visit ÷ first-timers in M | Person-level check-in export (exact); a bounded two-condition People list is the approximation | Monthly, one month lagged | None published anywhere |
| Giving participation rate | Active households with a gift in the quarter ÷ active households | Giving, By donor export keys matched against the active-household list | Quarterly | None; your own baseline |
| Giving concentration | Share of trailing-12-month giving from the top 10 households and top 10% | Giving, By donor export, sorted and cumulated | Quarterly | None; watch the movement |
| Volunteer participation | Adult attenders who served in the quarter ÷ adult attenders, intersected on Person ID | Services, with the Check-Ins integration on | Quarterly | 44% of participants (FACT) |
| Group participation rate | Adults on an active group roster who are also adult attenders ÷ adult attenders | Groups member export intersected with the attender export on Person ID; the Overview percentage beside it, on an undocumented denominator | Monthly | 44% of attenders (Lifeway) |
| Attendance year over year | This month vs same month last year; rolling 13-week average | Check-Ins event chart exported at 24 months (a 53-session report covers only the same-week comparison) | Monthly | Median church, 70 weekly |
The Benchmark column says “none” more often than not, and that’s the honest answer rather than a dodge. Size alone makes most comparisons meaningless: the median US church has 70 regular worship participants, while 78% of churchgoers attend the largest 13% of congregations. A “typical church” figure describes a church almost nobody is in.
One number to keep off the page: Total Donated from the Home dashboard widget. It includes donations that are uncommitted or not fully processed, so it moves between Tuesday and Thursday without a dollar changing hands.
What seven numbers still won’t tell you
They won’t tell you why. That part of the meeting needs people who know people, and no dashboard replaces it.
What they should do is survive the follow-up question. Somebody sees giving participation down two points and asks which households; if the answer is “I’ll look into that”, the pack did half its job. That gap is structural. Planning Center is a suite of separate products sharing a person record, so a report inside Giving can’t see Check-Ins, and the report that crosses two products is the one nobody can run.
It’s what we built Parable to close: every Planning Center product read nightly into one place where they share a person key, so the seven rates stay current and each clicks through to the households behind it while the board is still in the room, with SQL and warehouse access underneath.
If your board is content with a headcount and a giving total, none of this is urgent, and a second login to produce numbers nobody acts on is a step backwards.
But a board’s job is to see what a staff meeting can’t: direction, and risk, early. Both live in rates, and the rates live in data you already have.