Measuring a ministry without managing people

The first reaction a lot of ministry directors have to a dashboard is suspicion, and it is the correct reaction. Put a number in front of people and the number becomes the work. Count calls and you will get calls. Count closed cases and cases will close. Neither of those is the same thing as a family being looked after, and the gap between them is invisible precisely because the chart looks better than it did last quarter.

The worry underneath it is sharper than Goodhart's law, though. It is that measurement changes the thing you are looking at. A household in difficulty becomes a record in a status. The question stops being how that household is and becomes where it has got to. Nobody decides to make that trade — it arrives through the object they were handed to put a need in.

So this is not a guide arguing that more measurement is better. It is four tests a number has to pass before it has any business being on the screen of somebody who is about to phone a family, and a short list of things that should not be counted at all. They apply to a spreadsheet as much as to software.

One: it has to name somebody, not a rate

"Nine per cent of claims breached our turnaround commitment last month" is a report. It is useful to a board once a quarter and useless on a Tuesday morning, because there is nothing in it to do. Worse, a rate invites the response that improves the rate: the nine per cent gets reclassified, or the commitment gets loosened, and the families in it are exactly as stranded as they were.

"This claim is on day thirty-one, nobody has opened it, and here is the member's phone number" is work. A ministry number earns its place by ending in a person and an action. If the only available response to a figure is to feel better or worse about it, it is a board slide, and it should live on a board slide.

Two: it has to be arguable

Somebody who knows a family well has to be able to look at a score about them, disagree, and point at the specific reason they disagree with — not simply override it or learn to ignore it. That means the reasons and their weights are shown beside the number, and the whole rule set is readable, including the rules that did not fire.

A number that cannot be contested does not get trusted with pastoral care. It gets worked around, which is the same outcome as being wrong but harder to notice.

Three: a person has to be able to say "I have handled this"

Attention is finite and a list that only grows stops being read. So someone has to be able to clear an item — and what that must mean is "I have seen this and dealt with it", not "never show me this household again".

The distinction is load-bearing, and it is where most alerting gets it wrong in one direction or the other. An item that cannot be dismissed trains people to skim. An item dismissed permanently means somebody who has already decided a quiet member is fine will never be told that the same member is now quiet and in hospital. The resolution is that a dismissal lapses when the facts get materially worse, and nothing else brings it back.

Four: ties have to break toward the person, not the money

Any ranked list produces ties, and more of them than people expect — a bad week will put five households at the top with nothing to separate them. What happens next is a moral choice that usually gets made by accident, by whatever order the database returned.

Make it on purpose. In Auxilium, when two readings are equal, the one about someone hurting outranks the one about an expensive case. That is a line of code, and it is the kind of line worth knowing is there, because the default — process the biggest number first — is both defensible-sounding and exactly backwards for a ministry.

The same applies below the tie. If a board ranks by band but not within it, it is telling staff that five households are interchangeable, and they will work the top row. Whatever breaks that tie should be stated on the row itself, so the ranking can be disagreed with rather than merely obeyed.

What should deliberately not be counted

The tests above are about numbers worth having. This list is shorter and matters more.

No staff leaderboard. The moment contact volume is attributed and ranked, the cheapest contacts win and the long difficult conversation becomes a bad use of an afternoon. Follow-ups need an owner so that nothing is nobody's — an owner is not a score.

No close rate, and no target for one. A case closes when it is finished, and a number that rewards closing is a number that rewards closing early.

Nothing that scores a member on worthiness. Auxilium scores how well a household is being served and how defensibly a ministry is operating. There is no path anywhere in it from a score to a decline, and there should not be — a family refused by something that cannot be argued with has been refused twice.

And no fee per claim processed. That one is ours rather than yours, but it is the same principle: a vendor paid by the claim has a quiet interest in fewer of them, which would contradict everything else on this page.