When designing a metric framework, teams need to classify the a metric using a few questions.

• How close the metric does the metric tie to business success?
• What level of control do you have over a metric?
• How quickly will the metric move if action is taken to improve it?

Luckily, these three questions also break down into three categories of metric types, inputs, outputs and outcomes. By using this framework, teams can start to see what metrics they can directly effect and how those changes will propagate through their metrics framework to business success. Before we dive deeper into each, here’s a high level definition:

• Outcomes: Metrics that tie closest to business success but can only be moved indirectly. For example this would be revenue or customer retention which are key business results but usually cannot be acted on directly and take a long time to move consistently.
• Outputs: Metrics that result from inputs and will influence outcomes over time. These are typically faster moving, such as customer satisfaction scores from support surveys, and should be more directly influenced by changing inputs.
• Inputs: Metrics that can be directly affected by applying more resources. Going back to the support example, the time to first response is a metric that the team can directly impact by changing their processes or adding additional staffing.

Using this classification system, let’s dive deeper into how to identify the type of a metric and the implications it has on the decisions that can be made around it.

Outcomes

Outcome metrics are usually the highest-level, most directly tied to business success. As a result, they are often the place that most businesses start with their metrics frameworks. Typically there will be a revenue calculation before anything else get measured. Long-term business success requires that outcome metrics improve, such as increasing revenue, retention or profitability. Outcomes are the score card of a business and therefore get a lot of attention.

However, there are a few major points of caution when working with outcomes. They cannot be influenced in the short-term unless there’s a spike of resources that usually comes at a large cost, such as a temporary boost in marketing ad spend leading to increased customer acquisition. These temporary boosts usually feel great and can meet short-term needs but often the gains are also short-lived, with higher customer acquisition leading to higher churn because the customers acquired were not the right fit.

Furthermore teams have to recognize that they cannot directly influence outcome metrics. Without direct control, people in the organization, especially executives, can often feel a lack of control over the business. Not having control over outcomes is expected and is okay, it’s critical for people to recognize the limitations with outcome metrics. Data teams get stuck if they’re forced to analyze the outcome metrics for actionable patterns.

If your organization has been trying hard to find insights in outcome metrics, it may be time for you to go deeper into the framework and identify the outputs that could be influencing the outcomes and then the inputs that you can control.

Outputs

The difference between outcomes and outputs is commonly overlooked, along with most teams overlooking outputs in general. There are a few ways you can identify what’s an outcome vs what’s an output.

1. Is the metric able to change quickly? If the metric trend changes within a day or two of a project completing or a process change then it is most likely an output.
2. How closely tied is it to the input? If the metric can move as a direct byproduct of some input then it is most likely an output.
3. Is the metric a leading indicator? If the metric improving leads to another metric improving, it’s most likely an output metric improving an outcome metric.

Output metrics are not always a default part of the products and services companies produce. They often need to be created as ways to bridge the gap between inputs and outcomes. Many organizations fail to see the intermediary places they can measure and often end up with a framework closer to Input → Magic, Hopes and Dreams → Outcomes. The abstraction in the middle can be better defined and lead to much more actionable metric frameworks when built as outputs.

Let’s take a customer experience department as an example. One of the best leading indicators of customer retention is a highly satisfied customer. The department can create a customer satisfaction rating for each of the services they provide. The individual and overall ratings become output metrics that the team can use to judge the impact of their influence on inputs.

If you'd like help applying the input -> output -> outcome framework to your metrics, you can book a 30 minute session with one of our cofounders here 🔗

Inputs

Inputs are the metrics are the ones that your team has direct influence over. They produce short-term outputs that lead to long-term outcomes. Because they can be directly influenced, they are important to monitor and improve as much as possible. It’s also important to use business observability tools to find the actionable insights on these metrics because they can be changed quickly.

Giving every team member inputs allows everyone in the business to have actionable metrics  they can be accountable for. However, the process of integrating all the systems and defining metrics for all the major functions can be time intensive. We recommend taking an approach that aligns the data team’s focus with the current business objectives.

Applying Inputs → Outputs → Outcomes to the Metrics Framework

There will probably be gaps in your Metrics Framework that you identify. These are great opportunities for the data team to help the business strategy by proposing and executing on projects to fill in the gaps and make data more actionable. These might be business systems to integrate and model (Inputs), or product systems that capture feedback (Outputs) or major influencers of business success (Outcomes). Keeping the causal framework in mind and recognizing the limitations in influencing some metrics will make teams feel a lot more control and empowered to take action on data.

‍