Most growing companies don't have a metrics problem. They have forty metrics and no clear answer to a simple question: if revenue is off this month, which number moved, and whose job was it? A KPI tree answers that. It takes the one result the company cares about most and breaks it down, branch by branch, into the numbers each team can change this week.
This guide explains how the structure works, how to build one in an afternoon, and what it looks like for a SaaS company, an agency and an IT services firm. A free template is at the end.
Revenue
New revenue
Qualified demos
Win rate
Average deal size
Expansion revenue
Upsell rate
Churned revenue
Churn rate
- North Star
- Output
- Input
Key takeaways
- A metric tree breaks one North Star metric into the output and input metrics that produce it, usually through simple math.
- Most companies need three or four levels. Stop when the next level down would be a task, not a metric.
- Give every node one named owner. Review input metrics weekly, output metrics monthly and the North Star quarterly.
- Link each metric to the process behind it, so a drop points to a step, not just a number.
What is a metric tree?
A metric tree is a diagram that shows how one top-level business result is produced by the metrics beneath it. Each branch is a component of the number above it, usually connected by simple math: addition, multiplication or a ratio. Read from the bottom up, the tree shows what drives the result. Read from the top down, it shows where to look when the result changes.
That's the whole idea. The value comes from two things the diagram forces you to do: write down how your business actually makes its number, and decide who is responsible for each part of it.
Metric tree, KPI tree or driver tree?
You'll see all three names, and in practice people use them interchangeably. The differences mostly come down to where the term originated.
| Term | Where you'll usually see it | Typical top of the tree |
|---|---|---|
| Metric tree | Product and data teams, analytics tools | A product or growth metric, such as weekly active users |
| KPI tree | Management and operations | A company KPI with a target, such as quarterly revenue |
| Driver tree / value driver tree | Finance, consulting, corporate strategy | Profit, return on investment or shareholder value |
The DuPont model, which splits return on investment into margin and asset turnover, is one of the oldest driver trees still taught in finance courses. One more naming trap: search for "metric tree" and you'll also find a computer science data structure with the same name. It has nothing to do with business metrics.
The three levels: North Star, output and input metrics
A useful tree has three kinds of metrics. Most companies need three or four levels in total, rarely more.
North Star metric
This sits at the top. It's the result you're accountable for over a quarter or a year: revenue, gross margin, net revenue retention. It moves slowly, and nobody can change it directly on a Tuesday afternoon.
Output metrics
These are the components that mathematically produce the North Star metric. If revenue equals new revenue plus expansion minus churn, those three are your output metrics. Each one usually has a natural owner, such as a head of sales or a customer success lead.
Input metrics: leading vs lagging indicators
Inputs are the activities a team controls week to week: demos booked, proposals sent, onboarding calls completed, tickets resolved within SLA. They are leading indicators because they move before the result does. The North Star and output metrics are lagging indicators: by the time they move, the work that caused the change was done weeks ago.
The practical rule: manage input metrics weekly, review output metrics monthly and judge the North Star quarterly. If you only watch the top of the tree, you learn about problems a month too late.
Manage input metrics weekly, review output metrics monthly and judge the North Star quarterly.
How to build a metric tree in 6 steps
You can build a first version in two or three hours with your leadership team and last quarter's numbers.
- Pick one North Star metric for the period. One. Put revenue, profit and customer satisfaction at the top and you have three trees, so draw three.
- Write the formula for it. For a subscription business, recurring revenue for the month is starting revenue plus new revenue plus expansion revenue, minus churned revenue. If you can't write the formula, the tree will turn into a list of loosely related charts.
- Break each term down until you reach something a team can change this week. New revenue becomes qualified demos × win rate × average deal size. Qualified demos becomes inbound leads × qualification rate, plus outbound meetings. Stop when the next level down would be a task, not a metric.
- Test the math with real numbers. Plug in last quarter's figures. If the branches don't add up to the top within a few percent, either the formula is wrong or a branch is missing.
- Give every node exactly one owner. A person, not a department. Shared ownership of a metric usually means nobody checks it.
- Set a review rhythm. Weekly for input metrics, monthly for output metrics, quarterly for the North Star. Put it in the calendar now, or the tree becomes a slide nobody opens after the offsite.
Examples by business type
The numbers below are illustrative, chosen to show how the math works. Replace them with your own.
SaaS: an ARR tree
Top: annual recurring revenue (ARR).
- New ARR = qualified demos × win rate × average contract value. With 40 qualified demos a month, a 25% win rate and a $12,000 average contract, that's 10 new customers and $120,000 of new ARR per month.
- Expansion ARR = customers eligible for an upgrade × upgrade rate × average upgrade value.
- Churned ARR = ARR at risk × churn rate. The leading inputs here are product activation in the first 30 days and the share of accounts with a completed onboarding.
Owners: head of sales for new ARR, customer success lead for expansion and churn, product lead for activation. If you run a SaaS company, here's how this connects to day-to-day SaaS operations management.
Agency: a gross margin tree
Top: gross margin = revenue minus delivery cost.
- Revenue = available hours × billable utilization × average effective hourly rate. A 12-person delivery team with 140 available hours each has 1,680 hours a month. At 70% utilization and an effective rate of €95, that's €111,720.
- Effective rate drops when work goes unbilled, so track unbilled hours per project and change requests approved vs delivered.
- Delivery cost = team cost plus freelancers plus tools.
The agency trap is that revenue can look healthy while margin erodes through scope creep. The tree makes that visible, because unbilled hours sit right under effective rate instead of hiding in timesheets. More on running this in an agency operations setup.
IT services and outsourcing: a delivery margin tree
Top: delivery margin per account or per month.
- Billable capacity = team size minus bench time minus internal work.
- Rework = hours spent fixing delivered work, which is pure cost.
- On-time delivery rate, which protects renewals and reduces penalty clauses.
For IT outsourcing teams and MSPs, the useful insight is usually bench time and rework, not headcount. See how this maps to professional services automation.
From metrics to accountability: assigning owners
A tree without owners is a nice diagram. A tree with owners is a management system.
The test is simple. Point at any node and ask: who will notice first if this number goes the wrong way, and what will they do about it? If the answer is "the CEO, in the monthly review," the tree hasn't done its job yet. The founder is still the only link between the numbers and the work.
A tree without owners is a nice diagram. A tree with owners is a management system.
Three things make ownership real:
- One name per node, visible to everyone who sees the tree.
- A status next to the value. "Win rate 22%" means little on its own. "Win rate 22%, target 25%, at risk" tells the owner and their manager what to discuss.
- A link from the metric to the process behind it. If win rate drops, the owner should be able to open the sales process and see which step changed: discovery calls, proposals or follow-ups.
That last point is where most metric trees stop. The numbers live in a dashboard and the work lives in a project tool, so nobody can see that a missed onboarding step last week is why churn risk went up this week.
Common mistakes
Too many levels. Past four levels, people stop reading the tree. If a branch needs more depth, give it its own tree.
Branches that correlate but don't compute. "Website traffic" under revenue feels right, but unless it's part of a formula (traffic × conversion rate × deal size), it's decoration.
Metrics without owners. Covered above, and still the most common failure.
Vanity inputs. Counting activities that don't move the output metric above them, such as emails sent when the real constraint is reply rate.
Mixed time periods. Weekly input metrics feeding a quarterly North Star is fine. A monthly output metric calculated over a different date range than its parent is not.
Building it once. The first version will be wrong somewhere. After one quarter, check whether the input metrics actually predicted the North Star, and fix the branches that didn't.
Free template to build your own
Download the KPI tree template: a Google Sheet with the three levels laid out, formula cells that check whether branches add up to the parent, and an owner and status column for every node. There's also a blank PDF version for workshops.
KPI tree template
Three levels, formula checks, owner and status for every node.
How to run your metric tree in Enforcium
A spreadsheet is a good start. It gets hard to maintain once a dozen people own different branches and the numbers update every week.
In Enforcium, the Metric Tree shows how company-level metrics break down into team metrics, and each metric has an owner and a progress status (on track, at risk, off track). Metrics connect to Goals, so you can see which objective a number supports, and to Processes, so a metric can be linked to the specific process steps that drive it. When a step needs action, the built-in Tasks module keeps the work in the same place as the metric it affects. Leadership sees the result in the Live Dashboard without waiting for the monthly report.
FAQ
It starts with one North Star metric at the top and splits it into the metrics that mathematically produce it. Each level below is more specific and easier to act on. When the top number changes, you follow the branches down to find which input moved.
The North Star metric at the top (the result you're accountable for), output metrics in the middle (the components of that result) and input metrics at the bottom (the weekly activities a team controls).
Very little in structure. "Driver tree" or "value driver tree" is more common in finance and consulting and often starts from profit or return on investment. "Metric tree" is more common in product and operations teams.
Three or four for most companies. With fewer, the bottom level is still too abstract to act on. With more, the tree gets hard to read and maintain.
The CEO or COO owns the North Star metric and the structure of the tree. Every node below has one named owner, usually the person who runs the team or process that moves that number.