Engagement scores are common in community software. A member gets a number, or a color, built from how often they log in, attend, give and post. When the number falls, the member shows up on a list. It sounds like exactly what a busy team needs.
In practice, the list often goes unread. Staff look at a 38 and don’t know what it means, whether it matters, or what to say. So they don’t call. The problem isn’t the data underneath. It’s that the output is a number when it should be a sentence.
Why scores are attractive
Scores are easy to build and easy to show. They fit in a column, sort neatly, and make a dashboard look finished. They promise one number to watch for each person, and they suggest that someone has already done the thinking.
They also hide a set of choices. Someone decided that a login is worth a certain amount, that attendance matters more than giving or less, and that a month away counts the same in August as in November. Those choices may be sensible. But your staff can’t see them, so they can’t question them.
Where a bare score fails the people using it
They can’t be explained. A staff member can’t tell a colleague why someone is at 38. If you can’t explain it, you can’t act on it with any confidence.
They hide what moved. A drop could mean someone stopped attending, stopped giving, or stopped opening the app. Each needs a different response, and the score doesn’t say which.
They compare people with everyone else. A member who comes monthly and always has may score lower than one who came weekly for a month and vanished. The second person is the one who needs you.
They invite ranking people. Once members have numbers, it’s a short step to treating the bottom of the list as less worth your time. For a church, a nonprofit or an association, that runs against the whole point.
They don’t say what to do. A number going down is a fact about the number. The staff member still has to work out the rest.
What a reason looks like
An explained flag says what changed, compared to what, and shows the records behind it. It compares the person with their own past, not with the average. And it suggests one next step.
| Aspect | A score | An explained flag |
|---|---|---|
| What it says | Maria: 38 (down from 71) | Maria came to every Sunday service from January to June. She has missed the last five, and her small group hasn’t met since July. |
| What moved it | Unknown without opening the model | Stated: attendance stopped, and so did her group |
| Where to check | Nowhere obvious | Links to her attendance and her group’s meetings |
| What to do | Decide for yourself | Suggested: a check-in from her group leader, drafted for approval |
| When it’s wrong | Hard to tell | Easy to see: “She told us she’s caring for her mother until October” |
The last row matters most. Every flag will sometimes be wrong. Maria may be caring for her mother, and her group leader already knows. With a reason in front of them, staff can see that in a second and move on. With a score, they either act on something false or learn to ignore the list. After a few weeks of that, nobody trusts it.
Staff act on what they can explain
The test of any flag is whether the person reading it will pick up the phone. They will if they know what to say. “I noticed you haven’t been in a few weeks, and I saw your group’s been on a break. How are you doing?” is a call anyone can make. “Your engagement score dropped” is not.
Reasons also let staff learn. After a month of reading explained flags, a team starts to see its own patterns: people drift when their group stops meeting, or after a move, or when they stop serving. That’s knowledge about your community that a score keeps hidden.
The Science sets out the research and practice we design around, including the early signs that someone is drifting. A reason is how those signs reach a person who can do something about them.
Counting is still fine
This isn’t an argument against measurement. Attendance, gifts, sign-ups and group meetings are exactly what an explained flag is built from. A score can be a useful way to sort a list, and totals and trends belong in the board report. The rule is about what you show a person and ask them to act on: never the number alone. The reason travels with it.
How to test any tool
- Pick any flagged person and ask why they were flagged. Can someone on your team answer in one sentence?
- Can you see the records behind the flag, without asking the vendor?
- Does the flag compare the person with their own past, or with everyone else?
- Does it suggest a next step, or only a color?
- When staff know the flag is wrong, can they say so and move on?
If the answer to the first question is no, the rest don’t matter much.
The Assistant in Socio Connect
The Assistant flags who may be drifting and why. Where the platform keeps a score, it never shows one without the reason behind it. The commitment, as written: Shows its reasons. Every name on the list comes with why it’s there: a first visit, three missed Sundays, a gift that stopped. No unexplained scores. The Assistant drafts and your team decides what’s sent. Every name comes with a reason, and nothing it reads is used to train AI models.
Explained flags feed the Monday briefing, where each one comes with a drafted next step for a person to approve. The AI Assistant page has the rest.