This page is a work in progress, part of a multi-year effort to capture and share learnings, frameworks, tools, and processes to run organizations. See Running Organizations for more.

Measurement helps you to understand what's happening in a business operation and to make sense of performance. Measurement is the key to effective business processes and helps create accountability in the organization. You can't consistently improve an operation without good metrics.

Companies that don't measure anything (except for financial metrics) tend to either be one of two types: 1) Unstructured, wild, and ruled by fiefdoms and politics, or 2) Micro-managed and run by tyrannical leaders. Good measurement helps you avoid both of those scenarios.

Measurement is an ongoing job - it requires that you test metrics, and change and improve them as your operation changes. We never perfect metrics, only improve them over time.

Objectives (OKRs/Rocks) are goal-setting tools. They help you create improvements and change. Objectives ensure ambition in an organization. Key Performance Indicators (KPIs) tell you about performance. Objectives are how you lead an organization to a brighter future while KPIs are the key accountability tool for managing the organization’s performance.

KPIs measure the steady state of the organization and are predictive of the overall performance of the business. If performance on a given KPI is weak, then an Objective/Goal/OKR/Rock can be created to improve performance in that area.

Objectives and KPIs work well together - Atlassian and Gitlab are two companies that use OKRs and KPIs together.

Lagging Indicators (also called output metrics) tell you about performance. Common Lagging Indicators are measures like revenue, profitability, market share, CSAT, NPS. These metrics are out of your direct control - you can't act on a revenue number. The work you did last week, last month, last quarter, and last year produce the revenue output.

Outcome measures tell how successful you were, and point at areas that you can improve. You need accurate and timely reporting on Lagging Indicators. They may signal that a change in process is needed, a product isn't selling, or that quality is suffering.

Lagging Indicators don't tell why performance is good or bad, or how to change to improve. Lagging Indicators are commonplace and relatively easy to measure, which causes many organizations to rely too much on them.

Leading Indicators are also called lead measures or input metrics. Leading Indicators predict performance. Creating good Leading Indicators is one of the most high-leverage activities that management can take on because they tell you how to improve.

Leading Indicators need to be both causal (rather than correlative) and goal-based. Accomplishing those two things usually requires some finesse. It's fine to track "sales outreach," but what you really want to track is "high-quality sales outreach that leads to moving customers 1 step down the funnel."

Good Leading Indicators measure the success of your processes rather than the outcomes they produce. The more upstream, activity-based metrics you can track the better, assuming they're truly predictive of performance.

Leading Indicators can require a lot of legwork, and ongoing manual work. You may need Managers to observe a situation and document a result, rather than using automation or more objective measures.

Sean J. Taylor refers to the iterative process of Leading Indicator design in his article Designing and evaluating metrics. He explains that "the "lifecycle of a metric" involves an ongoing process of designing a metric, discussing it, validating it, experimenting with using it, optimizing it, and then changing it out when it's no longer helpful in predicting performance.

Graphic from Designing and evaluating metrics:

Finding a balance of Lagging and Leading Indicators is important for understanding the health and performance of an organization and continuing to create, test, refine, and change Leading Indicators is a key challenge for management.

North Star Metrics are singular Lagging Indicators the entire organization rallies around, usually in the pursuit of growth. A North Star metric makes simple and clear what the organization is optimizing for, above all else.

North Star Metrics can be user metrics, customer satisfaction metrics, operational metrics, or financial metrics. These metrics look a lot like BHAGs framed as KPIs.

Facebook's North Star Metric was Monthly Active Users, Slack's was Daily Active Users, Stripe's was processing volume. Frank Slootman used a North Star Metric at Service Now:

A good North Star Metric should represent sustainable business growth, and not be subject to "hacks" that can artificially inflate numbers. Using supporting and pairing metrics around a North Star Metric can tell you if that metric is being gamed or artificially inflated.

Early-stage companies should stay away from optimizing for single metrics and instead focus on producing a valuable customer experience. The customer experience at an early stage is likely a set of qualitative measures.

Whether you optimize to a North Star Metric or not, your organization should have 5-15 KPIs that are monitored on a regular basis. Common organization-level KPIs are metrics like Revenue, Cash Balance, Sales, Customer Satisfaction, Accounts Receivable/Payable, and Client Project Status. There’s a balance to strike, as the more metrics you monitor, the more noise and complexity you create.

Each organization-level KPI should have an accountable person assigned to it. Each metric should have a department or group of departments that are accountable for the performance of that metric.

KPIs should be balanced across the critical areas of the business. Many organizations overly rely on financial lagging indicators, which don't tell you anything about how your people are performing, whether customers are happy, and whether your processes are working or not.

Balancing your KPIs across finance, product quality, customer satisfaction, critical processes, and your various functions gives you a well-rounded view of the business.

KPIs should be designed to be updated and reviewed on a regular basis. Most organizations review their organization-level KPIs weekly, at the leadership team level.

There are many ways to report on performance, but the easiest way for small organizations is with a simple spreadsheet.

Here's one example from Traction Tools software:

You can also build your metrics into your One-Page plan and report directly out of your Leadership Team tool. Drawback: If you do not share your One-Page Plan openly with the wider organization, you'll need to duplicate your reporting to share it with the broader team.

Each Department or Team should have a set of 5-15 KPIs that tell them about performance in their area. Every core process in the business should have an associated KPI, especially those processes that cross over functions/departments.

For example, HR may have Lagging and Leading Indicators that cover areas like Recruiting, Employee Performance, Employee Engagement, Employee Development, and DEI.

Each of the Organization-level KPIs is owned by an appropriate Department, but each Department has additional KPIs that are not Organization-level KPIs. Some Departments may have no Organization-level KPIs they're accountable for.

Departments should be responsible for managing their own KPIs and keeping them updated in a public, accessible place, like a Dashboard and/or within an internal Knowledge Base.

Much of the business wisdom says that individuals should all have a number that they can be measured on and assessed by on a regular basis. Unfortunately, for most knowledge work and management roles, we can't reduce performance to a single number. Once management reduces individuals to a single number, the organization starts to lose sight of critical, less measurable ideals like our core values and operating principles.

Having a balanced set of KPIs for given roles and teams is a worthy goal. One way to humanize this effort is to ask people to create their own numbers by which they'll measure themselves. Creating our own numbers, divorced from management incentives, can help us improve our own work and ensure we're accountable to the larger organization.

The metrics you track should be informed by your goals. Start with the key questions that you want answers to in the business.

It's seductive to start with what's easy to measure and hope to improve your metrics over time, but that strategy is often no better than measuring nothing.

Measure only what's necessary. Track only those metrics that are likely to serve you in the near future. There are unlimited metrics one can track, but only a handful of metrics can move the business forward.

Report on positive metrics, not negative metrics. We want to focus on winning, not avoiding losses. Metrics that are framed around "getting worse" - avoiding losing clients, or preventing employee churn - are not motivating. Measure customer retention, not churn.

Don't rely on a single metric to tell you the truth. Pair two metrics together to present a fuller picture of reality. Analyzing a number of sales outreach attempts may tell you something about effort, but combining that metric with funnel velocity will tell you if that activity is creating results. Other examples:

Once you've defined metrics, you'll develop a sense of what "good" and "bad" performance is. Once you know what "good," is, it's hard not to try to aim for better, and to apply a number to that. Instead of setting targets for individuals and teams, use upper and low ranges to define performance.

An upper range is a realistic ideal and a lower range is the lowest you'll go without needing something to change. Deciding on a range of performance rather than a target helps clarify boundaries and creates shared expectations.

If you're struggling to reduce key indicators to just a few, consider rolling up multiple measures into an index. An index is a weighted average of many metrics (i.e. the stock market index). An index will simplify complexity while still allowing you to dig into specific areas if scores are low.

Don't rely on your first stab at a given Leading Indicator. Use qualitative assessment and your own judgment to verify that the metric is predictive of the important Lagging Indicator. Assess whether that Indicator is producing unproductive behaviors or externalities that undermine performance.

Assign each metric to a single person who will report on the metric on a consistent basis. The accountable person is not responsible for performance - the team/department/organization is responsible - they’re simply accountable for reporting.

A metric like "Customer Retention" can be measured in many ways (Gross Retention? Net Retention?), may be pulled from multiple channels, and is the responsibility of many people.

What exactly are you measuring? How will that measurement be taken? Write a sentence about the measure and what exactly tracks.

With many channel options for many metrics, define where that metric will be sourced from, and how you will pull it consistently.

The goal at this stage is not to assign a metric to each person, or try to decide how to determine if they're performing, but to understand how business performance will be understood.

Once you're clear on what your overall Organizational KPIs are, you can tackle measurement at a departmental level.

Start in one department, or with one team or core process. You can choose an area that is performing poorly or an area that is most ready for change. Focus on designing great KPIs, testing them, and improving them over time, and then standardize that process before moving to the next area.

Leadership can contribute with thoughts on Leading vs. Lagging Indicators that could be helpful, but if Leadership creates KPIs for Departments, you won't generate the commitment or sense of team responsibility that you need. KPIs will evolve and change as your organization does, and you need individuals to take responsibility for their performance by testing, tweaking, and improving their KPIs over time.

Reporting on your KPIs in dashboards allows everyone in the organization to understand performance at a glance - a powerful tool for alignment and motivation. Dashboards visually represent a significant slice of the reality of performance in the organization. Dashboards fit into a single page, and they "travel" well. They tell you what you need to know, quickly, to support or check a narrative.

You don't need a sophisticated database, tools for ETL, and a data visualization setup to get started with dashboards.

The most simple dashboard is a daily or weekly email that shows performance with Red-Yellow-Green color-coding (to represent acceptable ranges), and a simple table.

A Google Sheets "database" with a few automations, combined with data visualization in Google Data Studio, is a 100% free way to start visualizing data.

Automate as much of the data as possible to keep your reporting process as lightweight as possible. If reporting is time-consuming and error-prone, people will learn to dread reporting and the organization will learn to stop trusting metrics over time.

As you automate, continue auditing your metrics - you don't want to go 6 months incorrectly tracking an organization-level metric.

Default to posting Dashboards publicly, as much as possible. Publish them in your public Knowledge Base so that they're centrally located. You want people to "keep score" and know how the organization is doing, how their team is doing, and how nearby teams are performing.

Measurement can be taken too far. An organization isn't a machine or a series of machines, and because of that, you'll never be able to measure everything that matters.

Metrics, even good ones, are always imperfect - just proxies for the real thing. Metric design is an attempt to represent the truth, but metrics never tell the whole truth. We can't reduce health, happiness, or quality down to a single metric, reliably.

The McNamara Fallacy is the assumption that things that can't be measured aren't important. We fall into this fallacy because it gives us comfort that we can control outcomes by what we can see, and what we can measure.

Good metrics point us in the right direction and are useful when combined with feedback, conversations, intuition, and judgment.

Goodhart’s Law tells us that "when a measure becomes a target, it ceases to be a good measure." When we take a measurement focus too far, we end up trying to optimize measures rather than trying to optimize for the actual outcomes we seek. Here are two examples of where targets backfired:

One of the major failures of MBOs is that Managers demanded that individual performance be reduced to a metric. Over time, those metrics became targets, and MBO organizations introduced perverse incentives to hit numbers. The moment we reduce people to numbers and then set targets for our people is the moment at which we lose sight of what's important.

Another example of metrics becoming targets is from public school testing. Psychologist and Social Scientist Donald T. Campbell wrote about how the introduction of standardized test scores as a proxy for learning caused schools and school districts to set targets for scores, which therefore caused them to teach for tests, rather than teach for learning (Source: Campbell's Law).

We like metrics because they're supposedly objective. But if we can't accurately measure ideals like quality, creativity, and innovation, what are we to do? More importantly, if you choose to manage your organization by numbers, what ideals will you sacrifice? How will your core values be reinforced? Who will you become?

Don’t let metrics prevent you from sense-making, or from developing intuition.

Lots of companies use NPS and CSAT scores as Outcome metrics. Qualitative data is important and can lead to key insights, but any data collected via survey and reduced to numbers should be looked at with suspicion.

Survey design is difficult. It's hard to create unbiased questions that don't lead people toward the answers you want to hear. It's hard to know how people will interpret questions, and you can't assume that all people will interpret questions the same way.

How many responses did you get and who responded? It's hard to know if survey results are representative or just represent a customer segment or a sub-culture.

Moment-in-time surveys give you a snapshot of a moment and are usually heavily biased toward recent events.

You can use open-ended questions instead of multiple-choice questions, but you'll likely get fewer responses.

We’re usually better off spending a couple of hours calling customers and asking them questions instead of relying on open-ended surveys.