← All briefs
OperationsHigh Privacy

Newcomer Path Builder

A personalized 3-step next-step path for every guest from a 1-minute intake — drafted by AI, sent by staff.

A personalized 3-step next-step path for every guest, generated from a 1-minute intake — drafted by AI, reviewed and sent by staff.

CategoryOperations
Church-health domainOpen the Front Door
Data-privacy tierHigh — "where you're hurting" is a sensitive pastoral disclosure
Mastery-ladder targetLevel 3 · Skills
Build stackA short form + Make.com + Gmail + Claude API; logged to Planning Center

The problem

A generic welcome card collects a name and triggers a one-size-fits-all form letter — exactly the assimilation gap Nieuwhof flags (about 70% of leaders lack an effective process). The research on retention points the other way: connection that is relevant and relational is what holds people. A small staff can't hand-craft a tailored next-step path for every guest, every week — so everyone gets the generic email, and the generic email is the leak.

What good looks like

The welcome card becomes a 1-minute form: life stage, what brought you, and (optionally) where you're hurting. On submit, an AI reads the answers against the church's real ministry catalog and drafts a warm, specific three-step path — e.g., DivorceCare group → coffee with a deacon → a low-pressure newcomer dinner — as a Gmail draft. Staff review, personalize, and send. The chosen path type is logged, and the church reports 90-day return rate by path type.

Honest framing: no published study proves AI-personalized paths beat human follow-up. The bet rests on the well-sourced finding that early, specific connection predicts retention (McIntosh & Arn; Gallup), applied at a scale a small staff cannot sustain manually. The AI scales the drafting; the human keeps the pastoral judgment.

Market scan

ToolFitPricing (verify)
Text In ChurchTemplated, rules-based follow-up$31–81/mo
ClearstreamGuest texting sequencesCredit-based
Mailchimp / CRM automationsSegment-based sequencesVaries
SubsplashApp + sequences~$300–600/mo loaded
GlooAI-assisted engagementUnverified

The gap: templated tools branch on checkboxes; none publicly reads a free-text pastoral disclosure ("going through a divorce, new to town") and composes a bespoke path. That generative, staff-reviewed draft is the open lane — and the human-in-the-loop design is the whole safety model.

Data privacy & security

This is a high-risk front-door build, because "where you're hurting" can be health, grief, marital, or financial disclosure. Treat it like confidential pastoral care:

  • Explicit, plain-language consent on the form stating how the response is used and who sees it.
  • Store responses in Planning Center with restricted field-level permissions — never a public sheet.
  • Human-in-the-loop is mandatory: the AI only ever produces a Gmail draft. No automated sends, ever.
  • Don't feed raw PII to consumer AI — use the Claude API under Commercial Terms and minimize what you send (you can omit the full legal name).
  • Avoid logging the rawest disclosures in Make.com execution logs; provide an opt-out and a deletion path.

How to build it

  1. Build the intake form (Google Forms, Typeform, or Planning Center Forms) with a consent checkbox.
  2. Make.com watches for submissions.
  3. Make calls the Claude API with a tightly scoped prompt: "Given this life stage / motivation / need, propose a warm, pastoral 3-step path using only our real ministries [list]; do not promise counseling; flag anything needing urgent pastoral attention."
  4. Make creates a Gmail draft from the connections pastor — never auto-send.
  5. Staff reviews, edits, sends; Make logs the chosen path type to a Planning Center custom field.
  6. Weekly, compute 90-day return rate by path type (reuse the Guest-to-Member Funnel pipeline).

Rollout plan

  • Q1: Replace the welcome card with the consented intake form. Draft paths manually for a month to learn the patterns.
  • Q2: Turn on the AI-draft step with staff review; track path types.
  • Q3: Compare 90-day return rates across path types; double down on what works.

Effort & cost

  • Build: ~15–25 hours (prompt-tuning and the Gmail-draft step are the fiddly parts); ~5–15 min staff review per guest.
  • Run: ~$20–30/mo (form tool + Make.com $12–21 + a few dollars of API) plus review time.

Sources

  • Carey Nieuwhof — https://careynieuwhof.com/how-to-a-lose-first-time-guest-in-10-minutes-or-less/
  • Friendship factor (McIntosh & Arn) — https://www.apostolic.edu/the-friendship-factor-2/
  • Gallup — https://news.gallup.com/poll/16006/friendship-feeds-flock.aspx
  • Make.com pricing — https://www.make.com/en/pricing
  • Claude API data retention — https://platform.claude.com/docs/en/manage-claude/api-and-data-retention

Honesty flag: no study shows AI-personalized paths outperform human follow-up; the value case rests on scaling early, specific connection.