Designing Metric Trees

In this section, we discuss how you design a metric tree: by defining North Star metrics, recursively breaking down outputs into inputs, layering inputs, and adding metadata to understand influences.

Defining your North Star metric

Which metric(s) is/are most important – and why? It’s not an easy question to answer, but it’s a key question. Without choosing specific North Star metrics, your organization may end up dealing with a soup of metrics and no guidance.

What is a North Star metric?

North Star metrics are the top-level, high-order metrics in a company. Just as the phrase suggests, these are the metrics your organization should head toward. Another way to look at it is that North Star metrics are the metrics from which all other metrics derive their legitimacy. If another metric does not point to a North Star metric, it's not a legitimate metric.

Examples of North Star metrics

There are three types of North Star metrics: customer value, financial and strategic. All are equally valid, but whether the metric is a North Star for your organization depends on the fingerprint of your business.

For example, for a hotel booking platform, the number of nights booked could be a North Star metric. A social media platform might pick daily active users as a North Star metric – but it might choose total engagement time instead, depending on business priorities. For Chat GPT, a customer value North Star might be the number of prompts that are successfully returned.

Financial North Star metrics are arguably a bit more clear-cut, with total revenue and profit margins both valid examples of North Star metrics.

How do you deal with conflicting metric priorities?

Sure, you may decide to choose more than one North Star metric, but the point is that some metrics must have priority over others, and ideally, there should be one primary goal. The reality is that most organizations would have a few of these metrics to look at.

That’s why you need to define a North Star metric, as this topmost metric will define everything that comes after.

It’s also worth considering that your North Star metric may change as our organization evolves. Your organization will mature, and the industry may change. It is perfectly okay to adjust your North Star metric if and when needed.

If you live in a perfect world, and everyone in your business recognizes and understands its one true North Star, then start by placing that metric at the top of the tree. If you live in a more typical world, you would start with a well-known metric that’s accepted to have a high degree of leverage on your company’s success.

However, for most organizations, the North Star metric isn’t a huge matter of debate. Right at the top of the metric sits a dollar-based metric – a revenue outcome.

Decomposing Components

Let’s say we’re working on the metric tree for a marketing department. A north star metric might be Customer Acquisition Cost (CAC). The formula for CAC is Sales and Marketing Spend / New Customers, so we have two component inputs: Sales and Marketing Spend and New Customers.

That line of reasoning is a first step in decomposing metrics, but decomposing component metrics into a tree is a recursive exercise, so we’ll continue by breaking down Sales and Marketing Spend and New Customers into their component metrics.

Example: Sales and Marketing Spend

 The Sales and Marketing Spend component is composed of three input metrics:

  • Sales Spend: The sum spent on sales activities on closing Sales Opportunities in the period
  • Sales Development Spend: The sum of fully-loaded spend on sales activities on converting Prospects into Sales Qualified Leads in the period
  • Marketing Spend: Total spend on acquisition marketing in the period, inclusive of both variable marketing campaign costs and fixed costs e.g. salaries.

{Sales and Marketing Spend} = {Sales Spend} + {Sales Development Spend} + {Marketing Spend}

At this point, you might recognize the algebra. As we go down the tree, instead of simplifying the expression for CAC, we’re factoring it. So we could express CAC as the following:

CAC = ({Sales Spend} + {Sales Development Spend} + {Marketing Spend}) / {New Customers}

It illustrates the wonder in metric trees: once you’ve defined the relationships between your metrics and know your tree’s bottom-most inputs, you get definitive values for your entire tree.

Example: New Customers

The other component of CAC is New Customers, which we define as “The count of Customers whose first Subscriptions began in the period.”

In decomposing inputs, there’s a tough choice to make. Just how much detail do you want to go into? 

New Customers is a good example of where you have a choice in whether you want to decompose inputs further. You could choose to leave New Customers as an atomic metric—simply the count of new customers acquired in a period, or you could continue to break this metric down into further components.

If, for example, you chose to define New Customers as Closed New Business Opportunities x New Business Opportunity Win Rate, you would decompose it to a deeper level and expand the scope of the metric tree.

It would then encompass everything from the cost components to the productivity of lead acquisition efforts and the efficiency of sales efforts. You might then consider whether to express the whole lead funnel leading up to Closed New Business Opportunities as a tree composed of steps and conversion rates.

The takeaway is that your component definitions determine the shape of your tree, which determines the scope of your tree. Ultimately, the design choices you make when building your metric tree should lead to a metric tree that exposes the most actionable levers.

Identifying influences

Inputs and outputs are like the scaffolding of the metric tree. Now that you’ve established the scaffolding of your metric tree, it’s time to bring it to life with influence metrics. Unlike the rigid and predictable relationships between component metrics, influences are how metric trees account for the dynamism and uncertainty of a living business.

Influence metrics are also what make metric trees actionable. They reflect the levers operators can pull to “influence” component metrics that ladder up to a North Star metric.

Cost components are the exception here, but when it comes to growth, a lever’s leverage is rarely certain and is usually subject to diminishing returns. It would just be too easy if we could improve CAC by pulling the magical “Customers” lever.

As you decide what influences to hang from component metrics, remember that the relationship between an influence and an output may be well-accepted within your organization, or it may be hypothetical.

Since influence metrics are about action, you’ll want to focus on influences you actively or foreseeably manage or test. Beyond that, too many influences can lead to metric fatigue and ultimately water down the significance of a metric tree.

Win rate as an example

Let’s continue with our SaaS CAC example. Of the two components that make up CAC, the cost-related components are well-defined and easily understood. There isn’t any need to layer on a lot of nuanced influences.

As for the acquisition aspect of New Customers, there is a lot of variability and room for experimentation. Let’s look at Win Rate.

Identifying influence metrics depends more on business analysis, reasoning, and the hunches of internal subject matter experts than your proficiency in algebra.

If you ask sales managers what they believe influences win rate, they will likely list several intangibles. They also may suggest quantitative factors such as a sales rep’s experience and the number of interactions the salesperson has with the prospect. You might also find that the number of website visits correlates with whether or not a prospect becomes a customer.

Once you identify an influence, you’ll evaluate its utility and merit in belonging on the tree. There are two factors you’ll want to consider with each influence metric. They are strength and confidence. These factors are also encoded as metadata about these relationships.

  • Strength expresses the magnitude of the effect a change in an input will have on an output metric.
  • Confidence expresses the degree to which we are certain that a change in an input metric will result in a change in an output metric.

It’s worth noting that there are also practical considerations in choosing influences. Some influences will be more dynamic or less dynamic than others. That is to say, even if one metric seems to influence another, if that metric is impossible to move, it is probably not worth measuring.

Ready to build your own? Start with our Metric Tree Canvas template.

PAGE NAVIGATION