A data committee that works has an agenda, an owner, and a metric

The data committee is one of the first artifacts a data governance program creates, and one of the first to die. The pattern of its death is predictable: a packed, enthusiastic first meeting, a second with half the invitees, a third postponed, a fourth that never gets scheduled.

The cause isn't a lack of sponsorship. It's that the meeting doesn't decide anything.

Why it dies

A committee that only receives status presentations is expendable, because status can be an email. People with full calendars notice this quickly and hand their seat to someone with no authority, and then the committee loses the one thing that justified it.

The second reason is an agenda with no owner. "Discuss customer data quality" isn't an agenda item, it's a topic. Nobody prepares, nobody decides, everybody has an opinion.

The third is the absence of numbers. Without a metric, the discussion goes back to square one at every meeting, because there's no way to know whether anything improved.

The format that survives

Three blocks, one hour, always in the same order.

First: the metrics for the active domains, with the change since the last meeting. Five minutes, no presentation. The numbers were sent beforehand, and whoever didn't read them doesn't ask.

Second: the pending decisions. Each one with an owner, with the alternatives written out, and with the recommendation of whoever prepared it. The committee decides, it doesn't study. An item that arrives without written alternatives goes to the next meeting.

Third: the conflicts between areas that the steward couldn't arbitrate. This is where the committee earns its reason to exist: it's the only forum with the authority to say who wins when two areas disagree on the rule for the same data.

Who sits at the table

Whoever has the authority to change processes in their area. Not the technical specialist, not the representative. If the person has to check with someone else before deciding, the committee piles up pending items instead of resolving them.

That implies a small committee. Six people who decide are worth more than twenty who watch, and every extra invitation is the start of the hollowing-out.

What to record

Long minutes don't get read. The useful record has three columns: what was decided, who answered for it, when it gets revisited.

That record is the committee's most valuable asset. When the manager changes (and they do), it's what keeps the discussion from starting over from zero. A governance program with no memory of its decisions turns into a cycle of reinvention.

Cadence

Monthly works for most. Weekly wears people out and empties the room; quarterly loses touch with the operation and becomes a ritual.

If there's no pending decision and no conflict to arbitrate, cancel the meeting and say why. A committee that meets just to honor the calendar teaches everyone that attendance is optional, and that lesson is irreversible.