KPI Meaning: What a Key Performance Indicator Is, With Examples
A KPI is not just any number on a dashboard. It is the handful of measures an organization has agreed will tell it whether a specific objective is being met.
By Capstera Business Architecture Staff · Updated
14 min read
KPI stands for key performance indicator: a measure an organization has chosen as the clearest sign of progress toward a specific objective. Each KPI has a definition, a target, an owner and a review rhythm. Every KPI is a metric, but most metrics are not KPIs; the word "key" is a decision about which few measures matter most.
Most organizations do not lack measures. Open the monthly operations pack in a typical company and you may find a hundred or more of them, color-coded, trending, carefully maintained, and mostly ignored. When everything is measured, nothing is singled out, and a meeting that reviews forty indicators tends to discuss the three that turned red and act on none of them. A key performance indicator is the opposite move. It says: of everything we could watch, these are the few that tell us whether we are achieving what we set out to achieve, and when one of them moves, someone is expected to act. Choosing them is a strategic decision, not a reporting exercise, which is why business architects, who connect strategy to what the business actually does, are often asked to help.
Key Takeaways
- A KPI is a metric that has been tied to an objective, given a target and an owner, and singled out for regular review.
- KPIs measure ongoing performance; OKR key results are time-bound targets for change; KRIs warn of rising risk; SLAs are commitments to a customer.
- Balance leading indicators, which give early warning, with lagging indicators, which confirm results.
- Define each KPI precisely: formula, data source, unit, frequency, owner, baseline and target. Ambiguous KPIs produce arguments, not decisions.
- Anchor KPIs to business capabilities and value stream stages, so each one has a clear owner and nothing is counted twice.
- When a measure becomes a target, people optimize the measure. Pair KPIs that pull against each other to keep them honest.
What Does KPI Mean?
KPI means key performance indicator. Each of the three words does work.
"Indicator" is the modest word, and the most important. A KPI indicates; it does not explain. A falling customer retention rate tells you something is wrong, not what. Its job is to direct attention to the place where investigation is needed.
"Performance" says what is being indicated: how well the organization, or a part of it, is doing something it intends to do. That requires an intent. A measure with no objective behind it, such as the number of meetings held or pages published, may be interesting, but it is not a performance indicator because there is no stated level of performance it is being compared against.
"Key" is the decision. Out of everything that could be measured, an organization chooses a few to steer by. That choice is what turns a metric into a KPI, and it is why a KPI comes with obligations that an ordinary metric does not have.
- An objective it serves, stated in words, such as "deliver what customers order, when we promised it".
- A precise definition: formula, unit, data source and how often it is measured.
- A baseline and a target, so the number means something on its own.
- An owner who is expected to act when it moves the wrong way.
- A place in a regular review, where it is discussed and decisions are taken.
KPI vs. Metric, OKR, KRI and SLA
Several neighboring terms are used loosely, and the confusion causes real problems: a team asked for KPIs delivers a list of metrics, or sets key results that are really ongoing measures.
The simplest way to keep them apart is by the question each one answers. A metric answers "how much, how often, how fast?" about anything. A KPI answers "are we achieving this objective?" on an ongoing basis. A key result, the second half of an OKR, answers "did we achieve the change we committed to this quarter?" A key risk indicator answers "is a risk getting more likely?" And a service level agreement answers "are we keeping our promise to a customer?" The same measure can play more than one role: on-time delivery can be a KPI for the logistics function and, with a different threshold, an SLA in a customer contract.
| Metric | KPI | OKR key result | KRI | SLA | |
|---|---|---|---|---|---|
| The question it answers | How much, how often, how fast? | Are we achieving this objective? | Did we make the change we committed to? | Is this risk becoming more likely? | Are we keeping our promise to the customer? |
| Time frame | Any | Ongoing, reviewed on a cycle | Time-bound, usually a quarter | Ongoing | For the life of the agreement |
| Has a target | Not necessarily | Yes | Yes, a stretch target | A threshold that triggers action | Yes, contractually |
| Typical example | Website visits | On-time in-full delivery rate | Raise on-time in-full from its current level to the target by quarter end | Share of suppliers in one country | Systems available for the agreed share of service hours |
| Who owns it | Whoever finds it useful | A named manager | The team that set the objective | A risk owner | The service provider |
What Makes a Good KPI?
A good KPI can be calculated by two people independently and come out the same. Most bad KPIs fail that test before they fail any other.
The familiar SMART test, which George T. Doran set out for management objectives in a 1981 Management Review article, is a reasonable start: specific, measurable, assignable, realistic and time-related, in Doran's original wording. For KPIs, it helps to go further and write a definition card for each one. Organizations that skip this step discover, usually in a tense meeting, that finance and operations have been calculating "on-time delivery" differently for a year.
The card does not need to be elaborate. It needs to settle the questions that otherwise get argued about: what exactly is counted, from which system, how often, who is accountable, and what level counts as good.
| Field | What it settles | Example |
|---|---|---|
| Name | What everyone calls it | On-time in-full (OTIF) delivery rate |
| Objective | Why it exists | Deliver what customers order, when we promised it |
| Formula | Exactly what is counted | Order lines delivered complete by the promised date ÷ all order lines due in the period |
| Unit and direction | How to read it | Percentage; higher is better |
| Data source | Which system is the truth | Order management system, promised date as confirmed to the customer |
| Frequency | How often it is calculated and reviewed | Weekly calculation, monthly review |
| Owner | Who acts when it moves | Head of fulfillment |
| Baseline and target | What good looks like | Measured over the last two quarters; target set from that baseline |
| Thresholds | When it triggers action | Amber and red levels, agreed in advance |
| Capability | Which part of the business it measures | Order Fulfillment |
Types of KPIs
Three classifications are worth knowing, because a good set of KPIs mixes types deliberately.
Leading and lagging. A lagging indicator records a result after the fact: revenue, customer retention, on-time delivery. A leading indicator moves earlier and predicts the result: sales pipeline coverage, the share of orders with a confirmed stock allocation, employee engagement. Lagging indicators tell you whether you succeeded; leading indicators give you time to do something about it. A set made only of lagging indicators is a rear-view mirror.
Input, process, output and outcome. Inputs are resources consumed, such as spend or staff hours. Process measures describe how work flows, such as cycle time or rework. Outputs count what was produced, such as claims settled. Outcomes describe the effect that mattered, such as customers retained. Outcome KPIs are the hardest to move directly and the most important to watch; process KPIs are where improvement teams can act.
The Balanced Scorecard perspectives. Robert Kaplan and David Norton introduced the Balanced Scorecard in a 1992 Harvard Business Review article, arguing that financial measures alone describe past performance and should be balanced with customer, internal process, and learning and growth measures. The four perspectives remain a practical check that a KPI set is not lopsided.
| Perspective | The question it asks | Example KPIs |
|---|---|---|
| Financial | How do we look to shareholders and funders? | Operating margin; cost to serve; return on capital |
| Customer | How do customers see us? | Customer retention rate; on-time in-full; complaint rate |
| Internal process | What must we excel at? | Order cycle time; first-pass yield; change failure rate |
| Learning and growth | Can we keep improving? | Critical roles with a ready successor; time to competence for new hires |
KPI Examples by Business Capability
KPIs are easiest to own, and hardest to double count, when each one is attached to the business capability whose performance it measures.
The examples below are standard measures with widely used definitions. Use them as a starting point, then write your own definition card for each one you adopt; the formula details, such as what counts as "on time" or which contacts count as "resolved", are where organizations legitimately differ.
| Business capability | KPI | How it is calculated | Type |
|---|---|---|---|
| Order Fulfillment | On-time in-full (OTIF) | Order lines delivered complete by the promised date ÷ order lines due | Lagging, outcome |
| Inventory Management | Inventory turnover | Cost of goods sold ÷ average inventory value for the period | Lagging, process |
| Customer Onboarding | Time to onboard | Elapsed days from complete application to active account | Lagging, process |
| Customer Service | First contact resolution rate | Contacts resolved without a follow-up ÷ all contacts | Leading for satisfaction |
| Customer Relationship Management | Customer retention rate | (Customers at period end − new customers acquired) ÷ customers at period start | Lagging, outcome |
| Accounts Receivable | Days sales outstanding (DSO) | Accounts receivable ÷ credit sales for the period × days in the period | Lagging, financial |
| Accounts Payable | Invoice processing cycle time | Elapsed days from invoice receipt to approval for payment | Process |
| Procurement | Spend under management | Spend through managed suppliers and contracts ÷ total addressable spend | Leading for cost |
| Talent Acquisition | Time to fill | Elapsed days from approved requisition to accepted offer | Process |
| Product Development | Time to market | Elapsed time from concept approval to launch | Outcome |
| IT Change Management | Change failure rate | Changes that cause an incident or rollback ÷ all changes deployed | Leading for stability |
| Claims Adjudication | First-pass rate | Claims decided without being returned for information ÷ all claims decided | Process |
How Do You Choose the Right KPIs?
Start from what the organization is trying to achieve, not from the data that happens to be available.
1. Write down the objective in words. "Improve customer experience" is too vague to measure; "customers get what they ordered, when we promised it" is not.
2. Find the capabilities that drive it. Ask which business capabilities have to perform well for the objective to be met. For the delivery objective: order management, inventory management, warehousing and transportation. A capability map makes this step quick and keeps it complete.
3. Choose one outcome KPI for the objective. The measure that most directly shows whether the objective is being met, here on-time in-full delivery.
4. Add one or two leading or process KPIs for the capabilities that drive it. Stock availability at order confirmation, pick accuracy, carrier on-time performance. These give the owners something they can act on before the outcome slips.
5. Add a counterweight. Every KPI can be gamed. If the target is speed, add a quality measure; if it is cost, add a service measure. A warehouse measured only on pick speed will pick faster and wrong.
6. Write the definition card and set a target from the baseline. Measure for a period before setting the target, so it reflects reality rather than ambition.
7. Put it in a review and give it an owner. A KPI that is not discussed somewhere, on a known rhythm, by someone who can act, is a metric with a better name.
8. Prune. Review the set once a year. Retire KPIs nobody has acted on, and resist adding a new one without removing an old one.
KPIs and Business Architecture
Business architecture gives KPIs something they usually lack: a stable structure to hang them on.
Organizational charts change every year or two; business capabilities change rarely. When KPIs are attached to capabilities rather than to departments, they survive reorganizations, they can be compared across business units that perform the same capability, and their ownership follows the capability's owner.
The capability map also exposes gaps and overlaps in a KPI set. Shade the map by whether each capability has a KPI, and two problems appear at once: capabilities that matter to the strategy but are not measured at all, and capabilities measured by three teams in three slightly different ways.
Value streams add the time dimension. A value stream, such as settling a claim or fulfilling an order, crosses several capabilities, and its stages show where work waits or loops back. Stage-level measures such as elapsed time and rework rate are often the leading indicators that predict the outcome KPI at the end of the stream.
Finally, the link to investment. A capability whose KPIs are below target and whose strategic importance is high is a candidate for investment, and the business case for that investment should name the KPI it will move. That is the thread from a dashboard to a funding decision, and it is what makes benefits realization measurable after the money is spent.
Go further
- From capabilities to KPIs — How to measure what your organization actually does, capability by capability.
- Value stream KPIs — What to measure along a value stream, and why stage measures predict outcomes.
- Balanced Scorecard vs. OKR — Two frameworks for turning strategy into measures, compared.
- Business case: how to write one — Tie each benefit you claim to the KPI it will move.
A Worked Example: A Distributor's Delivery KPIs
An illustrative example. The distributor and its situation are invented to show how the steps fit together.
A regional distributor of building supplies has a single headline complaint from its customers: orders arrive late or incomplete, and contractors lose a day on site waiting for the missing items. The monthly operations pack has more than sixty measures, and none of them is the one the customers are complaining about.
The team starts with the objective in words: contractors get the complete order on the day we promised. From the capability map, four capabilities drive it: Order Management, Inventory Management, Warehouse Operations and Transportation Management.
They choose on-time in-full as the outcome KPI and write its definition card. The argument they had been having for a year, about whether a delivery made at 6 p.m. on the promised day counts as on time, gets settled in one line: the promised date is the date confirmed to the customer, and any time that day counts. Measured for two months before setting a target, it is lower than anyone had assumed, and the gap between the best and worst depot is wide.
For the driving capabilities, they add three process KPIs: stock available at order confirmation, pick accuracy, and carrier on-time performance. Each has an owner. As a counterweight to the push for speed in the warehouse, pick accuracy stays on the same page as picking productivity, so neither can be improved at the expense of the other.
In the first monthly review, the leading indicators do what they are for. Stock availability at confirmation is the weakest of the three, concentrated in two product families. The conversation moves from "deliveries are late" to a specific replenishment problem with an owner, which is the kind of decision a KPI exists to prompt. Thirty-nine of the sixty old measures are retired at the next annual review because nobody could name a decision they had informed.
Common KPI Mistakes
Most KPI problems are not about choosing the wrong measure. They are about what happens to a measure once people are judged by it.
- Gaming the target. The anthropologist Marilyn Strathern summarized an observation by the economist Charles Goodhart as "when a measure becomes a target, it ceases to be a good measure." Pair KPIs that pull against each other, and watch for improvements that are too sudden to be real.
- Too many KPIs. If a review covers forty indicators, none of them is key. Keep the set small enough to discuss.
- Measuring activity instead of results. Calls made, reports produced and meetings held are easy to count and rarely the point.
- No definition card. Two teams calculating the same KPI differently will spend their meetings reconciling numbers instead of improving them.
- No owner, or an owner who cannot act. A KPI owned by a committee, or by someone without authority over the capability it measures, will be reported and not managed.
- Targets set before the baseline. A target picked from ambition rather than measurement is either impossible or trivial, and people work out which within a month.
- Only lagging indicators. By the time a quarterly outcome moves, it is too late to change the quarter.
Frequently Asked Questions
What does KPI mean?
KPI stands for key performance indicator: a measure an organization has chosen as the clearest sign of progress toward a specific objective. A KPI has a precise definition, a target, an owner and a regular review, which is what distinguishes it from an ordinary metric.
What is the difference between a KPI and a metric?
A metric is any measurement. A KPI is a metric that has been tied to an objective, given a target and an owner, and singled out as one of the few measures the organization steers by. Every KPI is a metric; most metrics are not KPIs.
What is the difference between a KPI and an OKR?
A KPI monitors ongoing performance against a target. An OKR sets an objective and a few time-bound key results describing a change to be achieved, usually within a quarter. KPIs often show where an OKR is needed, and the KPI then monitors that the improvement holds.
What are some examples of KPIs?
On-time in-full delivery rate, customer retention rate, days sales outstanding, first contact resolution rate, inventory turnover, time to fill open roles, and change failure rate. Each should be attached to the business capability it measures and defined precisely before it is used.
What is the difference between leading and lagging KPIs?
A lagging KPI records a result after it has happened, such as revenue or customer retention. A leading KPI moves earlier and predicts the result, such as sales pipeline coverage or stock availability at order confirmation. A useful set includes both.
How many KPIs should an organization have?
As few as can be reviewed and acted on. A common practice is one outcome KPI per objective, supported by one or two leading or process KPIs for the capabilities that drive it. If a review cannot discuss each KPI, there are too many.
What is a KPI in business architecture?
A measure attached to a business capability or a value stream stage. Anchoring KPIs to capabilities gives each one a stable owner, survives reorganizations, exposes capabilities that are not measured, and links performance gaps to investment decisions.
What is the SMART test for KPIs?
SMART comes from George T. Doran's 1981 Management Review article on writing management objectives: specific, measurable, assignable, realistic and time-related. Later versions often substitute achievable and relevant. It is a useful check, but a full definition card with formula, data source, owner and thresholds settles more.
Pro Tips
- Shade your capability map by whether each capability has a KPI. Unmeasured capabilities that matter to the strategy will stand out immediately.
- Write the formula in words and as a calculation. The words settle the arguments; the calculation settles the spreadsheet.
- Put each KPI and its counterweight on the same page of the review pack, so neither is discussed without the other.
- Before adding a KPI, name the decision it will inform and who will make it. If you cannot, it is a metric.
- Retire one KPI for every one you add. The set only stays key if it stays small.