Kill your dashboard

Deep-dive lesson · about 10 minutes · short quiz at the end

ccPlanning academy · metrics · deep dive

Kill your dashboard

Most dashboards have forty numbers and drive zero decisions. Fix that.

The provocation

A full dashboard feels rigorous. It usually isn’t.

Forty tiles, every colour of the rainbow, refreshing every minute — it looks like control. In practice nobody knows which number to look at first, so they look at none of them, or the one that confirms what they already believe.

The test

Two questions per metric.

For every number on the wall, ask: what decision does this change, and who makes it? If you can’t answer both, the metric isn’t informing anything — it’s decoration. Most dashboards fail this test on most of their tiles.

Vanity metrics

Numbers that go up and feel nice.

“Total calls handled this month.” “Hours logged in.” They’re always reported, always grow, and never change a decision. Vanity metrics make a deck look busy and a quarter look productive while telling you nothing actionable.

Decision metrics

Numbers that change what you do.

A queue that triggers a real-time action. A repeat-contact rate that reprioritises a fix. A forecast-bias trend that changes the model. Each is tied to a specific decision and owner — that’s what earns a place on the board.

The sacred cows

Be willing to challenge the famous ones.

80/20 service level, raw AHT targets, occupancy maximised, login-time “productivity,” gross calls handled — the most-displayed metrics are often the least decision-useful, or actively drive bad behaviour. Tradition is not a reason to keep a tile.

Audience layering

Different roles, different boards.

The real-time analyst, the team leader, the planner and the director need different numbers at different cadences. One mega-dashboard serving everyone serves no one. Build a small, role-specific view for each — not a single wall of everything.

How to cut

Start from zero, earn each tile back.

Don’t prune the existing board — you’ll defend habits. Start with a blank dashboard and add a metric only when someone names the decision and owner it serves. Whatever doesn’t earn its way back was never helping.

The balance

Keep a few honest lagging outcomes.

Killing vanity metrics doesn’t mean keeping only live levers. Retain a small set of true outcome measures — resolution, a fair experience metric, cost — so you can still tell whether all the decisions added up to a better operation. Lean, not blind.

Run the test on one tile

“Calls handled this month: 48,210”

It’s on every monthly deck. So ask the two questions: what decision does it change, and who makes it? Nobody staffs, coaches or reprioritises off the gross total — it only ever goes up, and it’s big because the month was long.

No decision, no owner → it’s decoration. Kill the tile. Most boards lose half their squares to that one question.

The takeaway

Every tile earns its place or goes.

Ask what decision each metric changes and who owns it. Cut the vanity numbers and the sacred cows that fail the test, layer boards by role, and rebuild from zero. A small dashboard people act on beats a big one they ignore.

That’s the track — now test yourself ↓

1 / 10

Slides done? Here’s the same idea in a bit more depth — the part worth keeping.

In depth: forty numbers, zero decisions

A full dashboard feels rigorous — forty tiles, every colour of the rainbow, refreshing every minute — and it usually isn’t. In practice nobody knows which number to look at first, so they look at none of them, or only the one that confirms what they already believe. The cure isn’t a prettier dashboard; it’s a brutally smaller one where every tile has earned its place.

The two-question test

For every number on the wall, ask: what decision does this change, and who makes it? If you can’t answer both, the metric isn’t informing anything — it’s decoration, and most dashboards fail this test on most of their tiles. That separates vanity metrics — total calls handled, hours logged in, numbers that always grow and never change a decision — from decision metrics like a queue that triggers a real-time action, a repeat-contact rate that reprioritises a fix, or a forecast-bias trend that changes the model. And it means being willing to challenge the famous ones: 80/20 service level, raw AHT targets, maximised occupancy, login-time “productivity” — the most-displayed metrics are often the least decision-useful or actively drive bad behaviour. Tradition isn’t a reason to keep a tile.

Rebuild from zero, layered by role

The practical method is to start from a blank dashboard and add a metric only when someone names the decision and owner it serves — pruning the existing board just makes you defend habits, whereas earning each tile back exposes how much was never helping. Layer the result by role, because the real-time analyst, team leader, planner and director need different numbers at different cadences, and one mega-dashboard serving everyone serves no one. Just don’t over-correct: keep a small set of honest lagging outcomes — resolution, a fair experience measure, cost — so you can still tell whether all those decisions added up to a better operation. Lean, not blind.

The principle to remember: every tile earns its place or goes. Ask what decision each metric changes and who owns it, cut the vanity numbers and sacred cows that fail, layer boards by role, and rebuild from zero — a small dashboard people act on beats a big one they ignore.

Kill your dashboard