How to choose KPIs using a metric tree

The typical KPI selection process looks the same in almost every organisation. A leadership team gathers in a room, someone hands out Post-it notes, and everyone writes down the metrics they think matter. The group debates, negotiates, and eventually votes. The result is a list of 12 to 20 KPIs that reflect what people feel is important, not what actually drives the business.

This brainstorming approach fails for three reasons. First, it has no structure. Metrics from completely different levels of the business end up side by side, mixing strategic outcomes with operational activities. Revenue sits next to email open rate. Customer lifetime value sits next to support ticket response time. Without hierarchy, there is no way to see how these metrics relate to each other or which ones deserve the most attention.

Second, brainstormed KPIs lack causal logic. Nobody asks whether improving metric A actually moves metric B. The list is a collection of individually reasonable numbers that, taken together, tell no coherent story about how the business works. Teams end up optimising their own KPIs in isolation, sometimes at the expense of the broader outcome.

Third, the process rewards confidence over analysis. The loudest voices in the room tend to win, and the resulting KPIs reflect organisational politics rather than business reality. The metrics that survive the vote are the ones people already track, not necessarily the ones that would create the most value if improved.

The alternative to brainstorming is decomposition. Instead of asking "what should we measure?", you ask "what drives our most important outcome?" and break that outcome down into its component parts. Each level of the breakdown reveals the metrics that matter at that level, and the tree structure ensures every KPI is connected to the outcome it serves.

Start with your North Star metric, the single number that best captures the value your business creates. Then ask: what are the two to four factors that directly determine this number? Decompose each of those factors further, and keep going until you reach the operational levers that individual teams control. The KPIs do not need to be invented. They emerge from the structure of the tree itself.

This approach has a critical advantage over brainstorming: it produces KPIs that are connected by design. Every metric in the tree has a parent it feeds into and, in most cases, children that feed into it. When a KPI moves, you can trace the impact upward to see how it affects the business outcome. When the business outcome changes, you can trace downward to find the driver.

Decomposition also eliminates the problem of mixing levels. Executive KPIs sit at the top of the tree, department KPIs sit in the middle, and team KPIs sit at the leaves. There is no confusion about which metrics belong at which level, because the tree defines the hierarchy explicitly. The result is a set of KPIs that are not just individually good but collectively coherent.

Not every metric in the tree should become a KPI. A metric tree might contain 30 or 40 nodes, but only a subset of those deserve the focus, ownership, and reporting cadence that comes with KPI status. The following five criteria help you decide which metrics to elevate from the tree into your core KPI set.

1. 1

   Actionable

   A good KPI can be directly influenced by a team through their day-to-day work. If nobody in the organisation can take a specific action to move the number, it is a useful metric for context but a poor choice for a KPI. The metric tree helps here because the lower levels of the tree naturally correspond to things teams can control. If a metric sits high in the tree and no single team can move it, it is probably an outcome to track rather than a KPI to own.

2. 2

   Measurable

   The metric must be quantifiable and available on a regular cadence. Weekly or fortnightly is ideal for most operational KPIs. Monthly works for strategic KPIs at the top of the tree. If you can only measure something quarterly or annually, it is too slow to drive behaviour. A KPI that arrives after the window for action has closed is a report, not a performance indicator.

3. 3

   Connected

   This is where the metric tree earns its value. Every KPI should be linked to a parent outcome through the tree. If a metric cannot be placed in the hierarchy, if improving it does not demonstrably improve something above it, then it does not belong in your KPI set. Connectedness prevents the common failure of teams optimising metrics that feel important but have no traceable link to business outcomes.

4. 4

   Owned

   A KPI without a named owner is a number nobody is accountable for. Ownership means a specific person is responsible for monitoring the metric, investigating when it moves unexpectedly, and taking action to keep it on track. The metric tree makes ownership natural because each branch of the tree typically maps to a team or function. Assigning owners becomes a matter of matching the tree structure to the org chart.

5. 5

   Balanced

   Every KPI should be paired with a quality check that prevents gaming. If your KPI is lead volume, the balancing metric is lead quality or conversion rate. If your KPI is support ticket resolution time, the balancing metric is customer satisfaction score. Without a counterweight, teams will optimise the KPI in ways that damage the broader system. The metric tree makes these pairings visible because you can see sibling metrics on the same branch.

One of the most common questions in KPI selection is how many to have. The answer draws on decades of cognitive science research. Miller's Law, published in 1956, established that human working memory can hold roughly seven items, plus or minus two. Subsequent research has refined this further, suggesting that for complex information like business metrics, the effective limit is closer to four or five items before comprehension degrades.

The practical implication is straightforward: each team should own three to five KPIs. Fewer than three and you risk missing important dimensions of performance. More than five and you dilute focus to the point where nothing gets proper attention. When a team tracks 15 KPIs, they effectively track none, because no single metric receives the sustained attention needed to drive improvement.

Across the entire organisation, a well-structured metric tree typically yields 15 to 25 KPIs in total. This number feels large, but the tree structure makes it manageable because no single person or team is responsible for all of them. The executive team tracks the North Star and its three to four first-level drivers. Each department tracks its branch of the tree. Each team tracks the operational metrics at the leaves of their branch.

The tree structure itself prevents KPI proliferation. Every proposed KPI must justify its position in the hierarchy. If you cannot show where it connects to the tree, it does not make the cut. This is a far more disciplined approach than the brainstorming method, which tends to produce long lists because there is no structural constraint on what gets included. The tree acts as a natural filter, ensuring that every KPI has a reason to exist and a place in the broader system.

One of the most powerful properties of the metric tree is that it naturally segments KPIs by organisational level. Each tier of the tree corresponds to a different audience, a different time horizon, and a different type of decision. Understanding these differences is essential for setting KPIs that are appropriate for the people who will use them.

The tree ensures alignment without micro-management. An executive does not need to know how many demos a sales rep booked last Tuesday. But if new customer acquisition drops, they can trace down through the tree to find the level where the problem originated. Equally, the sales rep does not need to understand the full revenue decomposition. They need to know that their demo-to-close ratio feeds into new customer acquisition, which feeds into customer count, which feeds into revenue. The tree provides that line of sight.

This layered approach solves the common tension between leadership wanting visibility and teams wanting autonomy. Leadership gets visibility through the tree structure, which shows how every metric connects to outcomes. Teams get autonomy because they own the metrics at their level and decide how to improve them. The tree defines the what. Teams decide the how.

KPIs at each level also differ in their nature. Executive KPIs tend to be lagging indicators, outcomes that reflect past performance. Team KPIs tend to be leading indicators, inputs that predict future outcomes. The tree makes this distinction visible because lagging indicators sit higher in the hierarchy and leading indicators sit lower. This means the teams closest to the work are tracking the metrics that give the earliest signal of change, which is exactly where fast feedback loops create the most value.

Selecting the right KPIs is necessary but not sufficient. A well-chosen KPI without a system around it is just a number on a dashboard that nobody acts on. Each KPI needs four things to become operational: a target range, an owner, a review cadence, and linked actions.

The target range defines what good looks like. Avoid single-point targets where possible. A range, such as 72% to 78% for activation rate, acknowledges natural variation and prevents teams from overreacting to noise. The metric tree helps set realistic targets because you can model upward: if the team achieves 75% activation and the other branch achieves its target, what does that imply for the parent metric? Targets should be consistent across the tree, not set independently.

The review cadence determines how often the KPI gets formal attention. Match the cadence to the level in the tree. Executive KPIs reviewed monthly. Department KPIs reviewed fortnightly. Team KPIs reviewed weekly. Without a rhythm, reviews happen only when something goes wrong, which means you are always reacting and never anticipating.

Linked actions are what close the loop. When a KPI moves outside its target range, the owner investigates and decides on a response. That response is logged against the metric, tracked to completion, and measured for impact. Over time, this creates an organisational memory of what works and what does not. The metric tree provides the context that turns isolated KPIs into a connected system, so that an action on one metric can be evaluated for its impact on the metrics around it.