Leading vs Lagging Indicators: How to Find and Use Them

Most people learn about leading and lagging indicators as a simple distinction: lagging indicators tell you what happened, leading indicators tell you what’s likely to happen next. Use both, and you’ll have a more complete picture of performance.

That’s a reasonable starting point. But it leaves out the most important part — and in leaving it out, it leads to some persistent and costly mistakes in performance management practice.

A more useful perspective is this: leading and lagging are not fixed types of KPIs. They are roles that measures play in a cause-and-effect relationship that plays out with a time delay. Understanding that relationship — and how to find, validate, and use it — is what transforms performance measurement from a reporting exercise into a proactive performance management tool.

This guide covers what leading and lagging indicators actually are, why the relationship between them matters more than the labels, and the practical steps for finding lead indicators that have the predictive power you’re looking for.

What is a lagging indicator?

A lagging indicator is a measure of a result that has already happened. Sales revenue, client satisfaction scores, accident rates: all of these measure outcomes after the fact.

Despite this after-the-fact feedback, the strengths of lagging indicators are often underappreciated. They provide direct evidence of the results your organisation is trying to achieve — which is ultimately what performance measurement is for. And their signals over time, tracked in a well-designed chart, help you learn when, how and why their performance shifts. Historical analysis of lagging indicators provides the context that makes it much easier to anticipate future shifts, and to find the drivers of those shifts.

What lagging indicators cannot always do on their own is give you enough time to act. By the time a lagging indicator signals that something has gone wrong — customer satisfaction has dropped, revenue has fallen, the accident rate has risen — the contributing factors have often been in play for weeks or months. It means you’re doing more reacting than responding to performance signals.

Here are three examples of lagging indicators across different organisational contexts:

• Proportion of young people maintaining stable housing 12 months after leaving a support program (not-for-profit / social services)
• Percentage of procurement decisions meeting value-for-money criteria, by quarter (government / public sector)
• Net Promoter Score among existing customers, by month (corporate)

Each of these measures a result that matters. None of them tells you what to do about it until after the result has already occurred.

In truth, we can only measure things after they’ve happened — because the evidence to quantify doesn’t exist until something has happened. Forecasting analysis of lagging indicators is one way to give us a window — but never with perfect vision — of what the future might look like for a result that matters. But we will see how true leading indicators can give us a window with much clearer vision.

FUNDAMENTAL CONCEPT:
A lagging indicator measures a result after it has happened. It provides direct evidence of what your organisation is achieving — but by itself, it gives you limited ability to predict or prevent poor performance before it occurs.

What is a leading indicator?

A leading indicator is a measure of a result that causally precedes another result — and crucially, precedes it with a time delay. When the lead indicator changes, the lag indicator tends to follow, some time later.

That time delay is what gives leading indicators their value. Because the change in the lead indicator happens before the change in the lag indicator, you have a window to act — to reinforce what’s working, or to change course before the lag measure is affected.

But there is a distinction that most treatments of leading indicators miss entirely, and it matters enormously in practice: a leading indicator measures a result, not an activity.

This is where many organisations go wrong. A Health & Safety team counts the number of safety observations completed each week and calls it a leading indicator for accident rates. A sales team tracks the number of outbound calls made and calls it a leading indicator for revenue. Both are measuring activity — things that people did — rather than results that those activities were supposed to produce.

The best leading indicators measure the result of the activity, not the activity itself. The relevant question is not “how much did we do?” but “what changed because of what we did?” Safety culture (the degree to which staff internalise and practice safe behaviours) is a leading indicator for accident rates. The quality and relevance of sales conversations is a leading indicator for conversion. These are harder to measure — but they have actual predictive power.

A practitioner from a European mining company shared exactly this problem: their Health & Safety functions were recording strong performance on leading indicators — high counts of safety observations — but there was no corresponding improvement in the lag measures they were supposed to predict. The lead indicators looked healthy because they were measuring a lot of activity. However, they had no real causal relationship to the outcomes.

Here are three examples of genuine leading indicators:

• Proportion of case workers whose case plans are rated as strengths-based and goal-oriented, by quarter — leading indicator for client confidence and eventual client independence (not-for-profit)
• Attendance rate of small business advisory sessions, by quarter — leading indicator for 12-month small business survival rates
• Average depth of onboarding conversation rated by new customers, by month — leading indicator for 90-day retention (corporate)

Each of these measures a result — a change in the quality or character of something — not simply a count of activity.

FUNDAMENTAL CONCEPT:
A leading indicator measures a result that causally precedes another result, with a time delay. It is not an activity count. The time delay is what creates the advance warning — and the causal relationship is what gives it predictive power.

Leading vs lagging indicators: it’s a relationship, not a classification

Here is the most important thing to understand about leading and lagging indicators — and the thing that almost every explanation of them gets wrong:

Leading and lagging are not fixed properties of a measure. They are roles that a measure plays relative to another measure, in a specific cause-and-effect relationship.

The same measure can be a lagging indicator of one thing and a leading indicator of something else simultaneously. This is not a technicality. It is the central insight that makes performance measurement truly diagnostic rather than merely descriptive.

Consider three examples, each showing a measure in both roles:

Government / public sector: Staff Capability (proportion of staff meeting role competency standards) is a lag of Training Relevance Rating (proportion of training sessions rated by participants as relevant to their work roles). When training relevance improves, capability tends to follow — with a delay of months. But Staff Capability is simultaneously a lead of Overall Service Quality (average satisfaction rating by clients of overall service quality). When capability improves, service quality tends to follow — with its own delay. These three KPIs work in a lead-lag chain like this:

Training Relevance Rating → Staff Capability → Overall Service Quality

Corporate: Employee Engagement (proportion of staff rating their engagement as 8 or higher on a 10-point scale) is a lag of One-on-One Usefulness Rating (average rating by staff of the usefulness of their most recent one-on-one with their manager). But it is simultaneously a lead of Discretionary Effort Rate (proportion of staff reporting they regularly go beyond their minimum requirements). Engaged employees tend to be more motivated and productive — but it takes time to emerge. And these three KPIs work in a lead-lag chain like this:

One-on-One Usefulness Rating → Employee Engagement → Discretionary Effort Rate

Not-for-profit / social services: Client Confidence (average self-reported confidence rating at program midpoint) is a lag of Case Plan Ownership Rate (proportion of case plans where the client’s own goals are documented in the client’s own words). But it is simultaneously a lead of Independence Goals Achievement (percentage of client independence goals met at program exit). Confidence built during the program tends to predict the outcomes measured after it. Once again, three KPIs are working together in a lead-lag chain:

Case Plan Ownership Rate → Client Confidence → Independence Goals Achievement

In each case, the middle measure is neither purely a leading indicator nor purely a lagging indicator. Its role depends entirely on what it is being related to.

This means that asking “is this a leading or lagging indicator?” is often the wrong question. The more useful question is: “what does this measure lead, and what does it lag — and what is the evidence for each relationship?”

The practical implication is significant. Rather than choosing a set of leading indicators and a set of lagging indicators and treating them as two separate lists, the most useful approach is to map the causal chain — to understand how results at one level influence results at the next, all the way from the immediate results of the actions your organisation takes through to the ultimate outcomes it exists to create. This is what PuMP’s Results Map does, and we return to it in a later section.

The Southwest Airlines example

One of the most instructive real-world illustrations of a well-designed lead-lag chain — comes from Southwest Airlines’ Balanced Scorecard, documented by Robert Kaplan and David Norton in their landmark 1996 book, The Balanced Scorecard: Translating Strategy into Action.

Southwest identified gate turnaround time (the time between a plane arriving at the gate and departing again) as their central internal process measure. The lead-lag chain they mapped was:

Ground crew alignment → Fast ground turnaround → Flight is on time → More customers → Profitability

Turnaround time is a lag of ground crew alignment. It is simultaneously a lead of on-time arrival performance — which is itself a lead of customer volume and profitability.

When Southwest needed to reduce its fleet in its early years to survive financially, it was the turnaround time insight that made it possible. By reducing turnaround time to 10 minutes (down from the industry standard of 45–60 minutes), they could fly the same schedule with fewer aircraft. The lead indicator — turnaround time — was so well understood and so well validated against the lag measures it drove that the leadership team could make a bet-the-company decision based on it.

FUNDAMENTAL CONCEPT:
Leading and lagging are roles a measure plays relative to another measure, not permanent labels. The same measure is typically a lag of something upstream and a lead of something downstream. Understanding the chain is more powerful than categorising the measure.

Why leading and lagging indicators belong in a system, not a pair

Most organisations manage with a handful of KPIs that are loosely labelled as leading or lagging without a clear map of how they relate to each other. The problem with this approach is that it misses the diagnostic power that comes from understanding the full chain.

The Results Map — a core tool in the PuMP methodology — makes that chain explicit. It is a visual map of the cause-and-effect relationships between results across an organisation’s strategic and operational landscape, showing which results influence which others, in which direction, and at what level of the organisation each result sits.

Rather than a list of measures, a Results Map gives you a system of measures. Each measure has a position in the chain: upstream results that lead it, downstream results it leads, and the level of strategic or operational importance that determines how closely it needs to be monitored.

The fitness analogy

A useful illustration comes from personal fitness performance. Consider the Results Map for an athlete monitoring their fitness:

We can see potential lead-lag result chains throughout this map, and one of them is:

Body is well rested → Workouts are completed as planned → Training load grows but not too quickly or too slowly → Heart, legs and lungs are powerful

And the corresponding lead-lag indicator chain is:

Daily Readiness Score → Workout TSS → Weekly Training Stress Score (Weekly TSS) → Functional Threshold Power (FTP)

Weekly Training Stress Score — a commonly used measure of accumulated fitness — is a lag of Workout TSS and a lead of Functional Threshold Power. But it can be broken down further: Workout TSS is itself a lag of sleep quality, and FTP is a lead of other results like race performance.

Not every relationship in a Results Map is a lead-lag relationship, however. You still have to test for the time delay that a lead-lag relationship is defined by. But once you have, the map shows you where each indicator sits, what it predicts, and what predicts it. When Weekly TSS starts to drop, you don’t just note the number — you look upstream to find whether training quality has changed, and upstream of that to find whether the root cause lies in sleep or nutrition. Without the map, you’re responding to signals without context.

Applying the Results Map in organisations

The same logic applies to any organisational performance chain. A government agency trying to improve community wellbeing outcomes doesn’t just track the wellbeing measure — it maps the chain from the interventions it can directly influence, through the intermediate results those interventions produce, to the long-term outcomes that represent its ultimate purpose.

This is also why it’s important to monitor both lead and lag indicators — not just the lead. The lag indicator confirms whether the chain is working. If your lead indicators are improving but your lag indicators are not, the assumed causal relationship deserves scrutiny. Perhaps the time delay is longer than expected. Perhaps another factor is suppressing the outcome. Perhaps the lead indicator is measuring activity rather than a genuine upstream result. Without monitoring the lag, you cannot know.

FUNDAMENTAL CONCEPT:
Leading and lagging indicators are most powerful when understood as part of a system of cause-and-effect relationships — not as a pair. The Results Map makes that system visible, enabling diagnosis as well as monitoring.

The most common mistakes with leading and lagging indicators

Understanding the concepts is one thing. Applying them is where most organisations encounter difficulty. Here are the four mistakes that come up most consistently — and what to do instead.

1. Choosing leading indicators by intuition rather than evidence

A lead indicator is not declared — it is discovered. Intuition and experience are useful starting points for generating hypotheses, but the hypothesis has to be tested. An assumed lead-lag relationship with no empirical validation is just a belief. Beliefs sometimes turn out to be wrong in ways that cost organisations significant resources.

Fix: Before committing to a leading indicator, gather historical data for both the lead and the lag measure, and test for correlation and time delay. A scatter plot shows the strength of the relationship; a time series shows the delay. Both need to be present.

2. Ignoring lag indicators once lead indicators are in place

Once an organisation finds leading indicators it trusts, there can be a temptation to focus attention there and treat the lag measures as secondary. This is a mistake. Lag indicators are the measures of what you are ultimately trying to achieve. They confirm whether the lead-lag relationship is still holding — and alert you when it isn’t.

Fix: Monitor both. The relationship between your lead and lag indicators over time is itself a signal — if they start to diverge, something in the system has changed.

3. Treating the same measure as always leading or always lagging

As the examples in this guide illustrate, staff capability is a lag of training quality and a lead of service quality. Treating it as permanently one or the other means you will miss half of what it can tell you.

Fix: For each measure in your set, explicitly ask: what does this measure lead, and what does it lag? Map the chain in both directions.

4. Expecting immediate results from a leading indicator

If a lead indicator improves this month, the lag indicator it predicts will not necessarily improve next month. The time delay is part of what makes a lead indicator useful — but it also means that short-term impatience can cause organisations to abandon lead-lag thinking before it has had time to work.

Fix: When you validate a lead indicator, document the expected time delay and build that into your review cadence. A lead indicator for a result that takes two quarters to shift should be evaluated over at least two to four quarters, not one.

FUNDAMENTAL CONCEPT:
The power of leading indicators is in knowing that you’ve found the right ones. And this means avoiding a few mistakes in the process of discovering them.

How to identify leading indicators that actually work

Finding good leading indicators is genuinely difficult — more so than most guides suggest. It takes research, process mapping, and statistical testing, and it takes iteration. But the process is learnable, and the payoff is significant: a validated leading indicator gives you real advance warning and real ability to influence outcomes before they occur.

Here is the three-step process.

Step 1: Research

Before generating your own hypotheses, find out whether anyone else has already established relevant lead-lag relationships for the result you’re trying to predict. Academic research, industry benchmarks, and the performance measurement literature are all useful starting points. This step alone can save significant time — if others have already demonstrated that a particular lead indicator reliably predicts a specific lag result in your sector, you have a validated starting hypothesis rather than a blank page.

Step 2: Map the process

Flowchart the organisational process that your lag indicator relates to. For each significant step in that process, identify the result it produces — not the activity, but the outcome of the activity. Each of those upstream results is a candidate leading indicator.

For example, if your lag indicator is client outcomes at program exit, the process map might include: intake assessment quality → case plan quality → intervention quality → mid-program client confidence → exit outcomes. Each upstream result is a candidate lead indicator for the one downstream from it.

Step 3: Test statistically

Gather historical data for your candidate leading indicators and for the lag measure. Then test two things:

• Correlation — use a scatter plot to see whether the two measures move together. A tight, consistent pattern suggests a genuine relationship. A diffuse scatter suggests little relationship.
• Time delay — use time series plots of both measures to see whether shifts in the lead measure consistently precede shifts in the lag measure, and by how much. If the lead measure shifts first, and the lag measure follows by a consistent interval, you have a candidate worth tracking.

Where both conditions are present — strong correlation and consistent time delay — you have a validated lead indicator. Where only one condition is present, or neither, the assumed relationship needs revisiting.

The process takes time, and the first candidates you test may not hold up. That is normal. Finding powerful lead indicators is as much a practice of learning about your performance system as it is a measurement task — and the knowledge it generates is valuable regardless of which specific indicators you end up tracking.

FUNDAMENTAL CONCEPT:
A leading indicator is not chosen — it is discovered. Finding one that works requires research to identify candidates, process mapping to understand upstream results, and statistical testing to confirm the relationship and the time delay.