How One Mid-Sized Church Found 3 New Ministry Leaders in 6 Months Using AI Engagement Scoring

A 900-attendee church in the Midwest identified and developed three new ministry team leaders in six months by using AI-driven engagement scoring to surface members who were already behaving like leaders but had never been asked. Two now run small-group tracks. One took over the hospitality team. None of them were on the pastoral staff's radar in January.

Here is exactly how it happened.

The situation: a leadership bench that looked empty

Grace Community (name changed at the church's request) had the problem every growing church has. Sunday attendance hovered near 900. Small groups were capped because there were no new leaders to open more. The executive pastor, Marcus, kept a mental list of maybe six people he thought "might be ready someday," and every one of them was already leading something else.

The staff meeting refrain was familiar. We need leaders. We do not have leaders. Ask around and see if anyone comes to mind.

That method had produced two burned-out volunteers and one leadership candidate who quit after four weeks. Marcus was tired of asking the same ten people to do more.

What was tried: engagement scoring across the whole congregation

In late January, Grace connected its Planning Center data to ChurchAI and turned on engagement scoring across the full membership file, roughly 1,400 records including regular attenders who were not members.

The scoring model looks at behavior patterns that correlate with leadership readiness, not just attendance. It weights things like:

• Consistency of presence across services, groups, and serving windows over a rolling 90-day window
• Initiated contact, meaning the person reaches out first (replies to sermon follow-ups, asks questions, RSVPs early)
• Relational reach, or how often their name shows up connected to other members' first visits, group joins, or return after absence
• Giving trajectory, treated as a signal of ownership, not a filter (a $15 monthly recurring gift counts)
• Serving depth, not serving breadth: showing up reliably in one role beats sampling five

The system flagged 47 people scoring in the top decile who were not currently in any leadership role. Marcus expected to recognize most of them. He recognized 19.

What happened next: a shortlist, then conversations

The staff filtered the 47 down using a simple three-question review: Is this person theologically aligned? Have they been at Grace at least a year? Do we have any red flags in their pastoral care history? That left 22.

ChurchAI generated a one-page profile on each, drawing from communication history, group attendance, and serving records. Marcus and two other pastors read the profiles cold, without names attached at first, and each independently ranked their top eight.

Seven names appeared on all three lists. The staff invited those seven to coffee over the next three weeks. Not "come lead something." Just: "We have noticed how you show up. Tell us about that."

Three of those conversations moved forward within a month.

The three leaders, and what they had in common

• Dana, a 34-year-old accountant, had been at Grace eighteen months. She had quietly texted six different first-time visitors after their initial Sunday. The system caught it because those visitors returned at a rate well above baseline. She now leads a women's small-group cohort of 14.
• James, a retired teacher, showed up at every serve day and stayed late. Nothing flashy. His engagement score was driven by consistency and by the fact that four newer volunteers listed him as the reason they kept coming back. He runs hospitality on Sunday mornings, freeing a staff member for other work.
• Priya, 28, was almost invisible on paper. She rarely posted, rarely spoke up. But she had answered every sermon follow-up email in full paragraphs and had brought three friends into her small group, all of whom stayed. She now co-leads a young adults track.

None of the three had raised a hand. All three said yes when asked directly.

The result and the reusable takeaway

Six months in, Grace has three functioning ministry leaders, four more people in a formal development track, and a repeatable process the staff now runs quarterly. Marcus stopped keeping a mental list.

The takeaway is not that AI picks leaders. Humans still do the picking, and humans still do the developing. The takeaway is that most churches are already sitting on the data that reveals who is behaving like a leader, and that data goes unread because no staff has the hours to read it.

If your bench feels thin, it probably is not. It is probably just unsurfaced. That is the specific gap ChurchAI was built to close, alongside the disengagement and giving signals it also tracks. The product overview walks through the rest, and support can talk through what an engagement-scoring pilot would look like on your own data.

The leaders are already there. The question is whether you are looking at the signals that would show you.