Pastoral Care
Can AI help a pastor decide who to check in on?
AI can help a pastor decide who to check in on by surfacing the people whose participation has changed: the regular giver who stopped, the volunteer who stepped off a team, the group member who's been absent. It reads patterns across giving, serving, and groups that no one can hold in their head at scale. The pastoral decision of who needs a call, and why, stays with you.
I lead a church, and the hardest part of follow-up was never the conversation. It was knowing the conversation needed to happen at all. The people who are struggling are rarely the ones who tell you. They get quieter, show up less, and slide out of view while the loud needs in front of you take the week.
How do pastors know who to follow up with?
Most pastors know who to follow up with through memory, hallway conversations, and the people who happen to cross their path on a Sunday. That works for a congregation small enough to hold in your head. Past a couple hundred people, the quiet ones disappear, and the follow-up list becomes whoever you happened to notice this week.
That's the real constraint, and it has nothing to do with how much you care. A church of three hundred has thousands of small signals moving every month: gifts, serving slots, group check-ins, prayer requests. You see a sliver of them in person. The person who pulled back six weeks ago often isn't on anyone's list because the change was gradual and nobody was watching that particular thread. The tooling most churches run on stores all of that data but never connects it into a picture of one person.
Can AI tell a pastor who's pulling back from church?
AI can tell you who's pulling back by comparing each person's recent participation against their own normal, then flagging the ones who changed. If someone served twice a month and hasn't in two, or gave every month for two years and stopped, that shows up. It reads behavior across giving, serving, groups, and check-ins, never a single Sunday headcount.
The important word there is participation rather than attendance. No software watches the whole room on a Sunday and counts who's missing, and you should be skeptical of any tool that claims it does. What AI can read are the moments people leave a trace: a gift, a serving shift, a group sign-in, a form, a prayer request. I wrote more about that difference in participation vs. attendance, because it separates a tool that's honest about what it sees from one that's overselling.
What makes the flag useful is the math behind it. "At risk" as a number tells you nothing. The behaviors tell you everything. Compare these two ways of surfacing the same person:
| What the tool shows you | What you can do with it |
|---|---|
| "Sarah is a risk (score: 38)" | Not much. You don't know what changed or whether it matters. |
| "Sarah's giving stopped in March, her group check-ins have been spotty, and her last check-in was three months ago" | You can decide. That pattern reads like a life change, so you call her about her, instead of about a number. |
The second version respects your judgment because it shows you the same evidence you'd gather yourself if you had time to dig through every record. The first asks you to trust a label, and I'd never make a pastoral call off a label.
What can AI not see about your congregation?
AI can't see context, which is most of what matters pastorally. It can tell you a regular giver stopped in March and that their group check-ins went spotty around the same time. It cannot know that March was when the layoff happened, or the diagnosis, or the move to care for a parent two states away. That meaning is yours to learn.
This is why I don't think of any of this as automating pastoral care. It can't be automated, and it shouldn't be. What AI does is the part a human can't do well past a certain size: keep watch over hundreds of small changes at once and raise a hand when one of them shifts. The watching is mechanical. The caring is yours. When the software surfaces a name, the most common true story is something it could never have guessed, which is exactly why a person has to make the call.
There's a related question underneath this one about whether it's even right to point software at member data at all. I take it seriously, and I worked through where I land in is AI on member data ethical. The short version: it's appropriate when the software observes behavior and a pastor makes the judgment, and it crosses a line when it starts labeling people's faith or motives.
How do I keep the pastoral call in the pastor's hands?
You keep the call in your hands by treating AI as the thing that hands you a list, never the thing that decides what the list means. The pattern surfaces a name. You bring the relationship, the context, and the choice of whether and how to reach out. Used this way, the tool extends your attention without touching your discernment.
A few practical guardrails I'd hold to, whatever tool you use:
- Make it show its work. If a person is flagged, you should be able to see the specific behaviors behind it, not just a score. A flag you can't inspect is a flag you can't trust.
- Read for life events first. Most of the time someone pulls back, it's a season instead of a decision about your church. Walk in curious rather than corrective.
- Celebrate before you triage. Before you scan the list of who's pulling back, look at who's thriving and who's stepping up. People who lead from the deficit list burn out, and so do their congregations.
- Keep the relationship the point. The software's job ends when it puts a name in front of you. Yours starts there, and it looks like a text, a coffee, a real conversation.
If you want a fuller framework for sorting who actually needs a pastoral check-in versus who's fine, I wrote one in who needs a pastoral check-in. And if the people slipping past you are a recurring worry, keeping people from slipping through the cracks goes deeper on the systems side.
What does this actually look like in practice?
In practice, you open one screen on a Tuesday and it shows you the handful of people whose participation changed this week, with the specific reasons attached. You read the patterns, recognize the names, and decide which two or three deserve a real conversation. The software did the watching. You do the pastoring.
This is the one place I'll name what I'm building. Scout reads one connected record for each person, with their giving, serving, groups, check-ins, and notes in a single place, and writes a plain-language summary of who's pulling back and why, with the behaviors shown so you can judge for yourself. It carries a "Needs attention" status for people whose participation has changed, and it leans on celebrating who's thriving before it ever surfaces who's struggling. The manual version of all this works fine if you have the time to dig through records every week; Scout is the version that does the digging so you don't lose anyone to a busy month. What it will never do is make the call for you. I built it as a pastor who got tired of finding out too late. It was never meant to stand in for the part of the job that only a person can do.
Frequently asked questions
Can AI tell a pastor who to check in on?
AI can show you who has changed how they participate, like a regular giver who stopped or a volunteer who stepped off a team, by reading patterns across giving, serving, groups, and check-ins. It surfaces the names worth a second look. Deciding who actually needs a call, and why, stays with you.
Does AI replace pastoral discernment?
No. AI does the watching a human can't do at scale: tracking participation across hundreds of people week to week. It can't know that someone's quiet because of a new job instead of a hard season. You bring the relationship and the judgment. The software just makes sure no one falls off your radar by accident.
How does AI decide who's pulling back from church?
It compares each person's recent participation against their own normal. If someone gave monthly for two years and hasn't in three, or served twice a month and stopped, that change registers. Good tools show you the specific behaviors behind the flag instead of a vague risk score, so you can judge whether it matters.
Is it appropriate to use AI for pastoral follow-up?
It's appropriate when AI surfaces patterns and a person makes the call. The line to hold is observation versus judgment: software can notice that someone's participation changed, but it shouldn't label their faith or decide they're a problem. Used that way, it helps you reach people you'd otherwise miss.
What can't AI see about my congregation?
AI can't see context. It doesn't know about the diagnosis, the divorce, the deployment, or the season of caring for a parent. It sees that someone showed up less and can tell you when that started, but the meaning behind the change is yours to learn, usually over coffee.
Nic Moore is a pastor and the founder of Scout. The check-in I almost missed, and then didn't, is the reason this software exists.