The Weekly KPI Scorecard: 5 to 15 Numbers, a Free Template, Examples by Function, and an AI Pull That Works

By Andreas Pettersson, Founder, Leaders ADAPT

The numbers section was skipped. The data was not in yet. Thirty-six minutes later, the same team was designing a scorecard from scratch in the middle of issue solving: payroll spend, review count, occupancy.

Nobody owned the numbers, so nobody could read them.

That was a weekly meeting I sat in this summer. It is also the most common KPI scorecard failure I see. Not too few numbers. No owner for the ones you have.

This page shows you how to build a weekly scorecard that works, with examples by function, a free template and the one AI job that saves real hours. It is part of our guide to AI for EOS companies.

What goes on a weekly KPI scorecard, and how many numbers?

A KPI scorecard is a short weekly table of the numbers that predict whether the company will hit its plan, with one named owner and one goal per number.

Too many numbers is the issue. Almost never too few.

My rule is simple. Everyone needs one number they're measured by and one thing they can impact. A leadership team of seven people therefore starts with seven numbers, plus a handful the whole company watches. That lands most teams between 5 and 15.

If you have 40 numbers, you do not have a scorecard. You have a dashboard, and dashboards get scrolled, not discussed.

Three company numbers belong on every scorecard I have ever built: revenue, cost and cash position. Younger and smaller companies miss the cash position more often than you would think. They also fail to separate fixed cost from variable cost, so their margins swing with staffing cost, which is usually the bigger pool. Split them, and a bad month stops being a mystery.

Each row needs four things:

• The number, with a written definition of what counts.
• One owner, by name.
• A weekly goal.
• A 13 week view, so the trend is visible at a glance.

If you run EOS, the EOS scorecard is exactly this tool, read in the second slot of the Level 10 Meeting. If you run OKRs or Scaling Up, the same rules apply. The weekly numbers are your KPIs, and how they differ from objectives is covered in OKR vs KPI.

Department scorecards come later. Start with the company scorecard at the leadership table. Once it has run for a quarter without drifting, each leader builds a department scorecard underneath, and one number from each rolls up.

That is the cascade. It only works if the top one is trusted first.

Leading vs lagging indicators: which numbers belong on the scorecard?

A lagging indicator tells you what already happened. Revenue booked. Customers lost. Projects delivered late.

A leading indicator tells you what is about to happen. Qualified meetings held. Proposals sent. Tickets older than 48 hours.

I am a big fan of leading indicators, for one reason. The owner can still move them this week.

Revenue is not a scorecard number for your head of sales. It is a report card. Qualified first meetings held is the scorecard number, because it predicts the revenue and the owner controls it on a Tuesday afternoon.

Take a manufacturer that tracks on time shipments. By the time that number hits the report, the week is gone. The leading number underneath it might be orders released to the floor with complete materials. Put that one on the scorecard, and the shipment number mostly follows.

Keep a few lagging numbers for context, mostly the company three. Push the rest into a monthly review. The weekly scorecard is for the numbers you can still do something about.

Who owns each number, and what happens when it goes red?

Every number gets one ultimate owner. Not a team. Not "sales and marketing." One name. As the 5 Minute Leader puts it, "Co-ownership is non-ownership."

That takes very direct conversations about who owns what, and they are worth having before the number goes red, not after.

Then the weekly rhythm is short. In the L10 the owner says on track or off track. Off track moves to the issues list that week, no commentary.

Here is where most teams go soft. Someone says on track for everything, week after week, while everyone in the room knows it is not on track. The whole team has to challenge that out loud. Silence in that moment is how a scorecard dies.

When a number stays red for three weeks, stop discussing the number. Ask three questions instead:

• Is the goal wrong?
• Is this the wrong measure for the outcome we want?
• Does the owner have the authority and the time to move it?

The 5 Minute Leader has a line for the slow version of the same problem: "If a number has been yellow for three months, it's actually red."

Connect the numbers to the Rocks while you are at it. If a company Rock is to open a second location, one scorecard number should tell you every week whether that Rock is moving.

Expect to revise. An EOS implementer I spoke with put it plainly: a really great scorecard can take anywhere from a month to a year to develop. Your first version will be wrong in places.

That is fine. A wrong scorecard reviewed weekly beats a perfect one nobody reads.

KPI scorecard examples by function: services firm, agency, manufacturer, SaaS

These are illustrative starting points, one number per function, not benchmarks. Replace them with the number your owner can actually move.

The same method works below the leadership team. The implementer I mentioned scores even an executive assistant this way: one weekly yes or no on whether every client got everything they needed, on time and accurately.

Notice what repeats across all four columns. Cash position. Every business type runs out of it the same way.

A free KPI scorecard template (our own sheet)

Our weekly scorecard template is built on the one number per person rule. The layout looks like this:

If you have used a business scorecard template built around the balanced scorecard, notice the difference. A balanced scorecard is a strategy tool with a handful of perspectives and quarterly reviews. This is a weekly operating tool. Fewer numbers, faster review, one owner each.

Three rules for using it. Write the definition before the first week, or the number will drift. Change a goal only at a quarter boundary. Retire a number that nobody has discussed in a full quarter.

Where AI fits in the scorecard: the pull that works, and what stays human

The scorecard is the easiest place to start with AI in a company running EOS. It is also the easiest place to make a mess.

The pull that works is boring on purpose:

1. Connect the source systems: CRM, accounting, helpdesk, project tool.
2. Pull each number once a week at a fixed time, before the L10.
3. Compare it with the goal and flag every miss.
4. Write one line per red number saying what changed in the data, and nothing about why.

What breaks is predictable. Teams pull 60 numbers because the pull is free. Definitions drift, so "qualified" means one thing in March and another in May.

And AI starts writing explanations that sound confident and are wrong.

A lot of companies self-implement, do not put validation frameworks in place, and things go off the rails. Check the automated numbers against a manual count for the first month. After about a month of proper setup, with the right definitions and rules, it runs. Writing those definitions and checks before anything is automated is usually my first job as a fractional AI executive.

What stays human on the scorecard:

• Choosing which numbers matter, and retiring the ones that do not.
• The judgment about why a number moved.
• Who owns the number, and whether they stay in the seat.
• The consequence when it stays red.
• Goals the company commits to for the quarter.
• Any number reported outside, to a board, a bank or a buyer.

If you searched for an AI scorecard meaning a scorecard of your company's AI readiness, that is a different tool. Our AI readiness assessment answers that question. The line between automating and deciding is laid out in what not to delegate to AI, and the scorecard's place in the wider system is in the AI for EOS companies guide.

What a scorecard does not fix

The scorecard tells the truth. It cannot make anyone act on it.

It does not fix the owner who has been off track for six weeks and still says the plan is fine. It does not fix a leadership team that is too polite to challenge each other. It does not fix Rocks that were never connected to the numbers in the first place.

Those are leadership problems, not measurement problems.

EOS gives you a scorecard. It is lighter on the how: how to hold someone accountable for a number without it turning into blame, how to run the 1:1 where that conversation happens. That is the territory of our leadership team development guide. For the broader verdict on the system itself, see our independent review of EOS.

Growing people create growing organizations, and the opposite is also true. A scorecard just makes the difference visible sooner.

So here is my question. If you took your current KPI scorecard and crossed out every number without a single named owner, how many would be left? Send me the count.

Common questions about KPI scorecard

What is a KPI scorecard?

A KPI scorecard is a short table of the few numbers a leadership team reviews every week to see whether the business is on plan. Each number has one owner, a weekly goal and several weeks of history. In the Entrepreneurial Operating System it is called the Scorecard and it is reviewed in the weekly Level 10 Meeting.

How many numbers should a weekly scorecard have?

Most leadership teams do well with 5 to 15 numbers in total. A practical rule is one number per leader that the leader is measured by and can influence, plus a few company numbers such as revenue, cost and cash position. More than about 15 numbers usually turns a scorecard into a dashboard that nobody reads closely.

What is the difference between leading and lagging indicators?

A lagging indicator reports a result that has already happened, such as revenue booked or customers lost. A leading indicator measures an activity that predicts that result, such as qualified meetings held or proposals sent. Weekly scorecards lean on leading indicators because an owner can still change them before the quarter ends.

What happens when a scorecard number is red three weeks in a row?

A number that misses its goal goes to the issues list in the week it misses. If it stays red for three weeks, the team usually examines three things: whether the goal is realistic, whether the number is the right measure, and whether the owner has the authority and time to move it. The answer becomes a to-do with a date.

Can AI pull a KPI scorecard automatically?

Yes, if the numbers live in systems AI can read, such as a CRM, accounting software or a helpdesk. AI can pull each number weekly, compare it with the goal and flag misses. It should not explain why a number moved. That interpretation belongs to the owner, and the first month of automated pulls should be checked against manual counts.

What is an AI scorecard?

The phrase is used in two ways. Some people mean a KPI scorecard that AI fills in automatically from source systems. Others mean a scorecard that rates an organization's readiness for AI or the governance of its AI tools. The second kind is an assessment rather than a weekly management tool, and it answers a different question.

Your next question

• The Level 10 Meeting: where your scorecard gets read every week, in five minutes.
• The EOS Integrator: the seat that usually owns the scorecard process.
• Fractional AI executive: when you want the weekly KPI scorecard pull built and validated for you.