Cura — care
Hospitalization, bereavement, overdue pastoral follow-up. Somebody is carrying something heavy.
Narrative Relational Intelligence
Ministries fail members in a specific way: a family goes quiet, a case stalls in somebody's inbox, a promised follow-up never happens, and nobody notices until they leave. NRI notices.
Grace Okonkwo
Okonkwo household · primary contact
Someone is hurting
The case has stalled
Household complexity
Still in touch
Why 88, exactly
A score is the sum of the rules that matched. Add it up yourself.
Because those are the two things a spreadsheet cannot hold. A row tells you what a member's balance is today. It does not tell you the story that balance sits inside — that this household had a baby in March, a hospitalization in June, and has not replied to anyone since. Read as separate facts, none of those is alarming. Read as a narrative, it is the most important thing on your desk this morning.
And relational, because a ministry is not a claims processor with a warmer font. What is actually being tracked is the state of a relationship: whether it is being tended, whether it has gone quiet, whether a promise made in one conversation was kept in the next. Silence is the signal that matters most, and it is precisely the one that looks like nothing in a database.
So NRI reads the sequence rather than the snapshot, and it scores the relationship rather than the invoice. What it produces is not a risk score. It is a short answer to the question a staff member actually opens the software asking: who should I call today, and why them.
A member can carry several at once, and that combination is the useful part.
Hospitalization, bereavement, overdue pastoral follow-up. Somebody is carrying something heavy.
Large amounts, stalled processing, denials without a basis, coordination nobody owns.
Dependents, caregiving, a household in transition — a new baby, an adult child ageing off.
Unanswered outreach, incomplete onboarding, a relationship going quiet before renewal.
That is the entire algorithm. Each rule that matches contributes a stated number of points, the reasons are always shown beside the score, and they add up. A staff member who distrusts a number can check it by hand and either agree or point at the rule they disagree with.
There is no model, no training data, and no learned coefficient anywhere in the calculation. The full rule set is published to administrators inside the app — every code, every weight, and the reasoning behind each one. A system nobody can argue with does not get trusted with pastoral care, and it should not be.
When staff dismiss a signal it stays dismissed — until the facts get materially worse. Somebody who dismissed "this member has gone quiet" should absolutely be shown "this member has gone quiet and is now in hospital".
Five seeded members, one for each thing NRI is meant to catch, plus one who is perfectly fine.