Volunteer engagement comes down to six numbers: serving frequency, decline rate, declared availability, team tenure, the share of attenders who serve, and time to first serve. Planning Center Services records the raw material for all six. It reports almost none of them.
Key points
- Serving frequency is the metric that moves first, months before attendance or giving does. Count distinct weekends served this quarter against last quarter.
- Services holds the evidence: request statuses, decline reasons, blockout dates with a reason attached, per-team scheduling preferences, and attendance where a leader takes it.
- What it won’t do is hand any of it back as a spreadsheet: people reports print, matrix reports are PDF or a web page, and the CSV export is a roster of who is on your teams.
- Since March 2026 a volunteer can set one team to “Once a quarter” or “Unavailable” without leaving it, which is the clearest step back anyone can take in your system and produces no alert at all.
- There is no trustworthy national benchmark for what share of a congregation should serve. The credible sources disagree by more than fifteen points. Build your own baseline and watch the direction.
I wrote a while back that volunteers don’t quit, they stop being available. That post is about the pattern. This one is about the arithmetic: what to count, what the formula is, where the data sits in Planning Center, and what the number has to be before it’s worth a phone call.
What is volunteer engagement measuring?
Not attitude. You can’t measure whether someone loves your church, and every attempt to score it produces a number nobody trusts enough to act on.
What you can measure is behaviour with a time dimension: how often a person serves, how often they say yes when asked, how much of the calendar they’ve marked as unavailable, and how long they’ve been doing it. Every one of those is a rate rather than an event, which is why a single Sunday tells you nothing and a quarter tells you almost everything.
The second thing is the unit. “We filled 180 volunteer slots last month” behaves exactly like a Sunday headcount: it holds steady while the people underneath it change, which is how totals hide turnover. Six of your longest-serving people can leave the rotation in a quarter without the number moving, provided six new ones arrived. Compute all six of these per person, then aggregate. Never the other way round.
The six volunteer engagement metrics
| Metric | Question it answers | Formula | Where the data sits |
|---|---|---|---|
| Serving frequency | Who is serving less than they used to? | Distinct weekends served ÷ weekends available, this period vs. last | Plan Teams tab; Check-Ins volunteer check-ins |
| Decline rate | Who is saying no more often? | Declines ÷ requests, per person per quarter | Scheduling request statuses on each plan |
| Declared availability | Who has told you they’re stepping back? | Blocked-out days per quarter, plus any team preference set to a lower frequency or Unavailable | Blockout dates; per-team scheduling preferences |
| Team tenure | How long do people last on this team? | Months from first serving date to exit (archived or last served) for leavers, to today for active; medians reported separately | Derived from serving history |
| Share of attenders serving | How wide is the base? | Attending adults who also served at least once in the quarter ÷ distinct adults who attended at least once | Services and Check-Ins, intersected on Person ID |
| Time to first serve | Is the base being replenished? | Share of a cohort that served at all within 180 days, and median days to first position actually served among those who did | People and Services together |
Serving frequency
The core one. Count the distinct weekends a person served, not the positions they filled, because someone covering both services on one Sunday made one commitment, not two.
Express it as a rate, weekends served divided by weekends available, so a quarter with a holiday weekend in it compares cleanly against one without. Then compare the last thirteen weeks against the thirteen before. Thirteen is long enough to survive a vacation and short enough that the answer is still worth acting on.
A monthly-rotation volunteer serves roughly three times a quarter. When they serve once, that is a 67% fall in their rhythm, and it is invisible in every total you produce.
Two things in Services touch this without answering it. When you’re scheduling a team manually, the number next to someone’s name shows their last scheduled plan, which is genuinely useful at the moment you’re filling a slot and useless as a report. And the matrix will generate a Timecard Report showing the total time each person spent serving over the matrix’s date range. It is the closest thing Planning Center has to a frequency report, delivered as a PDF.
One caveat that decides whether your numbers mean anything: scheduled is not served. Services will track which team members show up for their scheduled times if a leader takes attendance, and if nobody does, your frequency metric is measuring your scheduler’s intentions rather than your volunteers’ behaviour.
Decline rate
Declines per request, per person, per quarter. One decline is noise. Two or three in a row from someone previously consistent is a change in rhythm.
Services records the response, and a volunteer who declines can add a decline reason. The matrix has a Statuses of People Scheduled Report that breaks out who is confirmed, unconfirmed and declined across the plans in view.
Now the part that catches people out. Planning Center documents the matrix display rule plainly: confirmed people appear in bold, unreplied people grey with a question mark, and “if people have declined, their name will not show up at all.” The screen your scheduler reads every week is the one place the decline signal is deliberately removed, because the matrix exists to show the hole, not the person who left it.
That gap widened in May 2026, when Services added volunteer replacements: a declining volunteer picks their own cover from a filtered list of eligible teammates, and once they confirm it, the replacement joins the plan and the decline is finalised. Excellent for Sunday. It also means a fifth consecutive decline can now resolve itself before any staff member hears about it.
Declared availability
The most underrated signal in Planning Center, because it’s the one where the volunteer has told you something explicitly.
Two mechanisms record it. Blockout dates let a person mark when they’re unavailable, with a reason for the team leader, and they apply across the whole product, so blockouts will block out that time frame for every team in Services. So blocking out four weeks is a statement about someone’s whole life, not one team’s rota. Count blocked-out days per person per quarter and watch the change rather than the level.
Scheduling preferences are the newer half. A volunteer can set, per team, the maximum number of times a month they want to be scheduled and how often they’d prefer to be. Since March 2026 that includes “Once a quarter”, which Planning Center implements as at least twelve weeks between schedules, and “Unavailable” for a specific team without leaving it.
Read that as what it is. Someone moving a team from every-other-week to once-a-quarter has made a decision about their involvement in your church and recorded it in your database, in a dropdown, with no notification to anyone but the scheduler who bumps into the warning. Both reports Planning Center offers here describe the current state: a Blockout dates people report, and a matrix report of anyone who declined because of a blockout. Neither shows the change, and the change is the signal. Neither keeps it either. An expired blockout or a preference moved back up disappears from the next run, so the quarter-over-quarter comparison only exists if you export both reports on a fixed date each quarter and keep the files. A snapshot you didn’t take is a trend you can’t have. Quarterly snapshots see what was standing on the day. A two-week holiday blockout that came and went between them is invisible, and that is the point: this metric is about sustained declared availability, not every absence. If you want the within-quarter churn too, export monthly; the comparison stays quarter against quarter.
Team tenure
Median months on the team, per team: for someone who has left, first serving date to the date they were archived or last served; for someone still serving, first serving date to today. An active monthly volunteer’s tenure runs to the reporting date, not to their last Sunday, and it is still open. Report the two groups separately, because a median that mixes finished tenures with ones still running understates both. It is the metric that tells you whether your kids ministry is a place people stay or a place people cycle through, and it is the one your team leaders will argue with, usefully.
Nothing computes this for you. You derive it from serving history, which makes one administrative decision unexpectedly load-bearing: archiving a volunteer keeps their scheduling history, while removing them also removes them from past plans in which they were scheduled. Archive people who step back; remove only people who were never scheduled. A church that tidies up by removing lapsed volunteers deletes its own turnover data.
Share of attenders serving
Distinct people who served at least once in the quarter and appear among the distinct adults who attended at least once in the quarter, divided by those adults. The intersection, not the whole Services list: a youth volunteer or someone marked attended in Services who never checked in sits above the line with nothing below it, and the rate can pass 100%.
Both halves need care. The denominator only exists if you record adults by name. A sanctuary counted with Headcounts gives you a number and no people, which is the limit of what attendance reporting can give you. The numerator is not your Services team-member count, which is a roster of everyone with a profile in the product, including people who last served in 2023.
There’s a quieter distortion if you count volunteers through Check-Ins: a volunteer whose location doesn’t match a Services team or position name is recorded as Regular rather than Volunteer, making your serving number low and your attendance number high. One of several reasons check-in data needs reading carefully before anything gets built on it.
Time to first serve
Median days between someone’s first attendance and the first position they actually served. Attended, not merely scheduled, the same distinction serving frequency already makes. A newcomer scheduled in week three who declined and first showed up in week nine was replenished in week nine; counting the schedule date reports a first serve that hadn’t happened, or never did. Measure it per cohort. Everyone who first attended in January, then everyone who first attended in February. Report two numbers, not one: the share of the cohort that has served at all by a fixed horizon (say 180 days), and the median days among those who did. The median alone flatters a bad cohort: a group where three of forty served, fast, shows a quicker median than one where thirty served in three months, because the thirty-seven who never served have no date and drop out of the middle. Compare cohorts only once each has reached the horizon.
This is the only forward-looking metric of the six. The other five describe the volunteers you have; this one tells you whether the people arriving this year become the volunteers you have next year. It’s usually the first thing to break when a church gets busy, because welcoming and scheduling are run by different people who compare notes twice a year.
It also refuses to live in one product. First attendance is in Check-Ins, membership is in People, first position is in Services, and the Person ID is the join key. Same shape as the groups-to-serving pipeline in small group metrics, and one reason a church metrics dashboard has to be built across products rather than inside one.
What can Planning Center Services give you?
Everything above is recorded. Here’s what comes back out, and in what shape.
| Report | Where it lives | What it gives you | Format |
|---|---|---|---|
| Blockout dates | People page, print icon | Current blockouts per person | Print or PDF |
| Future schedule | People page, print icon | Dates people are scheduled for, ahead | Print or PDF |
| Statuses of People Scheduled | Matrix, print icon | Confirmed, unconfirmed and declined across the matrix | PDF or web page |
| Blocked Out People (open teams) | Matrix, print icon | Anyone who declined because of a blockout | PDF or web page |
| Timecard | Matrix, print icon | Total time each person served in the matrix’s range | PDF or web page |
| CHECK-INS: Absent People | Plan actions, with the integration on | Confirmed on the schedule, never checked in | PDF or web page |
| People page export | Services People page | Selected people’s information, scheduler permissions or higher | CSV |
Two patterns fall out of that table. Every report about serving behaviour is built to be generated and printed. A clipboard for a team leader, not a file for a pivot table. And the one thing that exports as CSV is the roster: who is on your teams, not when they served.
The other route is People lists, which filter people on their information and activity across all Planning Center applications, and Services attendance is available as a list condition. A list will answer who served between March and June and hasn’t since, as a yes or no. What it can’t hold is how often, because a list is a filter with a date on it rather than a history. That is the distinction behind every report Planning Center can’t run.
How do you build a serving frequency comparison?
Here’s the version that works, tedious parts included. Budget an afternoon the first time.
- Check the Check-Ins and Services integration is on and matching. A scheduled volunteer who checks in is marked attended in both products, which is what puts serving data somewhere exportable. Session times must match the plan and event location names must match your team or position names, or the rows land under the wrong type.
- Pick two equal windows. Thirteen weeks against the thirteen before. Write the exact dates down before you start; you’ll want them again next quarter.
- Build a custom report on your weekend event. On the event’s Reports tab, set the Layout to separate pages by check-in type so Volunteer is its own section, and download it as CSV rather than PDF.
- Check the shape of the file before you build anything on it. You need one row per person per date. If your export has collapsed to one row per person, the dates are gone, and you’ll need to run the report per session and stack the files.
- Keep the Person ID column and match on it. Never on name. Two Sarah Johnsons and a Mike who is Michael in half the rows will corrupt every number downstream, quietly.
- Add a week-number column derived from the check-in date, so two services on one Sunday count as one weekend.
- Pivot by person with two measures: distinct weekends served in the earlier window, and distinct weekends served in the recent one.
- Convert each to a rate and take the difference. Then filter to people who were actually regulars first. Someone who served twice in the earlier window can’t meaningfully drop, so set a floor of three or four weekends and sort by the size of the fall.
- Put blockouts and preferences beside the top twenty. Run the Blockout dates people report and check each name’s team preferences. A volunteer who set one team to “Once a quarter” in April isn’t drifting; they told you, and nobody read it.
- Cross-check against attendance and giving before anyone calls. Serving down with attendance down too is a different conversation from serving down and attendance unchanged, and a lapsed gift on top of both is a third conversation again.
If your church doesn’t check volunteers in, step 3 has no CSV to give you. The honest fallback is generating a People Calendar report from a matrix covering your date range as a web page and pasting it into a spreadsheet. A transcription job, not an export. Getting your Planning Center exports into a sheet on a schedule is worth doing before you attempt it a second time.
Is there a benchmark for how many of your people should serve?
Not a usable one, and I’d rather say so than repeat a number that sounds authoritative.
Look at what the credible sources actually report. Lifeway Research found that Protestant pastors estimate, on average, 42% of adults in their churches volunteer regularly, with 18% of pastors putting their own church under 20%, up from 13% of pastors saying the same in 2008. Ask churchgoers directly and you get a different picture: two in three say they have not volunteered for a charity, church included, in the previous year. The Hartford Institute’s survey of over 24,000 church attenders reports that half of attenders volunteer monthly. Gallup, measuring the whole population rather than congregations, has volunteering for a religious organization at 39%, up four points since 2021.
Those aren’t contradictions so much as four different questions: regularly versus at all, pastor’s estimate versus member’s self-report, congregation versus country. None of them settles what your church’s number should be.
The gap between the first two is the argument for measuring this at all. Pastors estimate 42%. Members report far less. Estimation is not a reliable instrument here, including yours.
While I’m at it: the 80/20 rule, where a fifth of the congregation does most of the work, is an aphorism rather than a finding for volunteering. It has real support on the giving side, where the concentration is well documented, but I could not locate a single measurement of it for serving from a research organisation, and Lifeway’s data points the other way, with more than half of pastors placing their church at 40% or higher. Don’t quote it in a board meeting.
So build your own baseline. Compute the six metrics for the last four quarters, write the numbers down, and treat those as the benchmark. Direction is the thing you can act on. A serving base that fell from 34% to 29% over a year is a real finding about your church; the same 29% next to somebody else’s national average is a debate about survey methodology.
What these numbers change
The point of the six is not a scorecard. It’s that each one converts into a list of names, which is the only test worth applying to any metric a board sees: when the number moves, can you say who moved it? A serving frequency chart that falls and can’t name a person is a chart to read aloud. The same chart with eleven names under it is a Tuesday morning.
Which is why this rarely survives as a spreadsheet habit. The afternoon isn’t the problem; the answer being stale in three weeks is, and the person who built it has the least free time. That is the specific job Parable does. It reads Services, Check-Ins, Giving, Groups and People nightly into one place where they share a person key, so serving frequency, declines and availability sit on the same row as attendance and giving, with SQL and warehouse access underneath.
Under a couple of hundred people you don’t need any of this. Your kids ministry director knows who’s wobbling, and she’s right more often than the data would be.
Past that, the schedule stops being something a person can hold in their head, and it turns into what it always was underneath: the earliest written record you have of who is quietly changing their mind about your church.