Designing a KPI scorecard your clients will actually read

Five to nine measures, one owner each, read weekly. How to pick them, what to cut, and the one number every owner is avoiding.

7 min read · All guides

Key takeaways

Five to nine measures. Below five misses a dimension of the business; above nine and nobody reads it weekly.
Every measure needs a goal, one named owner, and a weekly reading. Missing any of the three and it is decoration.
Prefer leading indicators someone can actually move over lagging ones nobody can influence this week.
The measure the owner keeps forgetting to update is usually the most important one on the list.

What a scorecard is for

A scorecard exists to start the conversation from facts. Read it at the top of every planning session and every check-in, before anyone offers a theory, and the quality of the discussion improves immediately.

It is not a dashboard and it is not management reporting. It is the short list of numbers that tell you whether the business is working, readable in ninety seconds by someone who was not there when it was built.

How many, and why

Five to nine. Fewer than five and you are almost certainly missing a dimension — usually the team-side or the customer-side one. More than nine and the weekly read becomes a chore, gets skipped, and a scorecard nobody reads is just a spreadsheet with feelings in it.

The constraint is doing real work: forcing the choice between two similar measures is often the most valuable half hour of the setup.

What to include

One cash measure. Days of cash, collections, or whatever version fits the business. Always.
One margin or profitability measure. Revenue alone hides too much.
One or two leading indicators. Things that move before revenue does — quotes out, pipeline created, bookings.
One customer-side measure. Retention, complaints, on-time delivery, response time.
One team-side measure. Open roles, utilisation, turnover, safety.
One quality measure. Rework, error rate, returns — whatever the business's version of doing it twice looks like.

That is six categories. Most good scorecards land between six and eight measures once duplicates are removed.

What to cut

Cut anything nobody has looked at in two quarters. Cut vanity measures that only ever go up — cumulative totals, all-time counts. Cut anything that requires someone to build a report to produce, because it will not be produced weekly.

And cut one of any two measures that say the same thing. Businesses routinely track revenue, bookings and invoiced value as three separate lines when two of them move together in every scenario that matters.

Goal, owner, reading

Every measure needs three attributes. A goal, so the reading means something. One named owner, so there is a person to ask. A weekly reading, so the trend is visible before the quarter ends.

A measure missing any of the three is decoration. In particular, an unowned measure is a measure nobody updates, and by week six it is stale enough that people stop trusting the whole scorecard — not just that line.

The number they are avoiding

Every owner has one. Gross margin by job. Effective hourly rate. Days sales outstanding. Churn. It is the number that would force an uncomfortable decision, and it is missing from the scorecard for reasons that are always framed as data-availability problems.

Getting it onto the scorecard is frequently the highest-value thing a coach does in the first quarter of an engagement. Not because measurement is magic, but because the avoidance was the actual issue, and the measure makes it undeniable.

Reading it well

Colour against goal, not against last week. Look at the trend before the level — a red number improving for four weeks is a different situation from a red number flat for four weeks, and treating them identically destroys credibility.

And when a measure is red twice in a row, it should generate an agenda item automatically. Otherwise the scorecard becomes a thing you look at rather than a thing that causes action.

Where QuarterOS fits

QuarterOS puts the scorecard at step two of the guided flow — the client's own measures, with goals, owners and weekly readings, coloured against target and carried into next quarter's look-back.

See the scorecard →

FAQ

Questions this raises.

Weekly or monthly readings?

Weekly where the number moves weekly, monthly where it genuinely does not. Do not force a weekly reading on a measure that only changes at month end — that is how scorecards acquire fake data.

Should the scorecard be the same across all my clients?

The structure should be; the measures should not. Consistent structure means you can read any client's scorecard instantly, and client-specific measures mean it reflects their actual business.

What if the data does not exist?

Start with a manual estimate and improve it. A rough weekly number someone actually looks at beats a precise number produced quarterly by a report nobody reads.