KPI (Key Performance Indicator)

Definition: A KPI (Key Performance Indicator) is a specific, measurable metric a team tracks because it directly signals progress toward a defined goal. It’s the general category of “the number that matters” — the underlying building block that more specific frameworks like North Star Metric and OKRs (Objectives and Key Results) are constructed from.

How It Works

What Makes a Metric a Good KPI

Not every number a dashboard can display deserves to be called a KPI. A genuinely useful KPI tends to share several traits:

Measurable — it can be captured consistently and objectively, with a clear method for calculating it the same way every time
Tied to a real goal — it connects directly to something the business actually needs to achieve, not just something that happens to be easy to track
Actionable — when the number moves, someone knows what levers to pull in response; a metric nobody can influence isn’t a useful KPI, just an observation
Owned by someone — a specific person or team is accountable for it moving in the right direction, rather than it floating as everyone’s and no one’s responsibility
Timely — it can be measured often enough to catch problems or confirm progress while there’s still time to act
A metric that fails several of these tests — hard to measure consistently, disconnected from a real goal, nobody accountable — is a candidate for the dashboard’s trash bin, not its centerpiece
Good KPIs are also comparable over time: a metric whose calculation method keeps changing can’t reliably show whether things are getting better or worse, even if each individual snapshot is accurate

Leading vs. Lagging KPIs

Lagging KPIs measure outcomes that have already happened — revenue closed, last month’s churn, customers who already canceled. They’re accurate and easy to defend, but by the time they move, the underlying cause is already in the past
Leading KPIs measure activity that predicts a future outcome — sales calls booked this week, this month’s product activation rate, trial-to-paid conversion so far. They give a team time to react before the lagging number arrives
A well-designed KPI dashboard pairs the two: a lagging KPI to confirm whether a goal was actually hit, and one or more leading KPIs upstream of it that the team can influence in real time
Leading indicators are inherently a bit noisier and less certain than lagging ones, since they’re a prediction rather than a confirmed result — which is exactly the tradeoff that makes them useful early-warning signals instead of after-the-fact scorecards
Most functions have at least one natural leading-lagging pair: sales calls booked leads to deals closed, trial activations leads to paid conversions, support first-response time leads to customer satisfaction scores

Vanity Metrics vs. True KPIs

This is one of the most important distinctions in choosing what to track at all:

A vanity metric goes up and to the right in a way that feels good to report, but doesn’t reliably connect to whether the business is actually getting healthier — total signups, app downloads, or social media followers are classic examples
A true KPI is tied to a real economic or strategic outcome — activated users, paying customers, revenue retained — even if the number is less flattering or grows more slowly
The test that separates the two: if this number doubled overnight, would anything about the business’s actual health necessarily improve? Downloads doubling with no change in active or paying users changes nothing real; paying customers doubling changes everything
Vanity metrics aren’t always useless — a rising signup count can be a genuine leading indicator further upstream — but treating them as the headline KPI, rather than a supporting data point behind a real one, misleads a team about how the business is actually doing
Founders are especially prone to reporting vanity metrics in pitch materials because they’re the biggest, most impressive-looking numbers available, which is exactly why experienced investors probe past them to the metrics underneath
A simple gut check: a vanity metric usually answers “how much activity happened,” while a true KPI answers “did that activity actually create value” — the second question is harder to answer but far more useful

KPI vs. North Star Metric vs. OKRs

	KPI	North Star Metric	OKR (Key Result)
Scope	Many, spread across functions	One, for the whole company	A handful per team, per cycle
Time horizon	Ongoing, no fixed end date	Ongoing, rarely changes	Time-boxed (usually quarterly)
Purpose	Ongoing health gauge for a specific area	Single unifying measure of value delivered	Specific, time-boxed target tied to a stated objective
Changes how often	Rarely — same KPIs tracked for quarters or years	Very rarely — only with strategy shifts	Every cycle, by design
Example	Support ticket resolution time	Weekly active teams collaborating on a document	“Grow weekly active teams from 10,000 to 14,000 this quarter”

Every team can have its own KPIs — support has resolution time, sales has quota attainment, product has activation rate — while a North Star Metric is deliberately singular, meant to align the entire company around one definition of value delivered
An OKR’s Key Result often is a KPI, just with a specific numeric target and a deadline attached — “reduce churn” is a KPI worth watching continuously, while “reduce monthly churn from 5% to 4% by end of Q3” is a Key Result built on top of that same KPI
KPIs don’t expire; a support team tracks resolution time indefinitely. OKRs are explicitly temporary, reset every cycle, and often retired once achieved or replaced by a new priority
In practice, a healthy company’s stack looks like: one North Star Metric at the top, a handful of KPIs per function feeding into it, and quarterly OKRs that call out which specific KPI needs deliberate, focused improvement right now
None of the three frameworks replaces the others — a company that only sets OKRs without any steady-state KPIs underneath them tends to lose sight of ongoing operational health in between planning cycles

Measuring KPI Attainment

Many KPIs are tracked against an explicit target, especially once they feed into an OKR cycle or a performance review:

\text{Attainment Rate} = \frac{\text{Actual Result}}{\text{Target}} \times 100

Worked example: a support team sets a KPI target of resolving 90 out of every 100 tickets within 24 hours (a 90% target). If they actually resolve 78 out of 100 tickets within that window, attainment comes to $\frac{78}{90} \times 100 \approx 86.7$ .

An attainment rate below 100 doesn’t automatically mean failure — it depends on how aggressively the target was set in the first place, which is itself a judgment call worth revisiting each cycle
Tracking attainment rate over several cycles, rather than a single period in isolation, reveals whether a team is consistently under- or over-shooting its own targets, which is useful information for calibrating future goals
Targets are typically set one of two ways: top-down, derived from a company-level goal such as an OKR or a board commitment, or bottom-up, built from historical performance plus a reasonable stretch
Overly conservative targets that get hit every single cycle stop functioning as a stretch goal, while chronically unreachable targets erode a team’s trust in the target-setting process itself

The SMART Criteria for KPI Targets

A common framework, borrowed from general goal-setting practice, for writing a KPI target well:

Specific: the target names an exact number and an exact metric, not a vague direction like “improve retention”
Measurable: the underlying data needed to check progress is actually available and trustworthy
Achievable: the target is a real stretch, not a number picked from thin air with no grounding in current performance
Relevant: the target connects to a goal the business actually cares about right now, not just a number that’s convenient to track
Time-bound: the target has a clear deadline or review date, so attainment can actually be assessed rather than tracked indefinitely with no checkpoint

Goodhart’s Law and KPI Gaming

Goodhart’s Law — often summarized as “when a measure becomes a target, it ceases to be a good measure” — is one of the biggest risks in any KPI system
A support team measured purely on ticket resolution time can hit an excellent number by closing tickets prematurely before issues are actually fixed, technically satisfying the KPI while making customers worse off
A sales team measured purely on new deals closed can be incentivized to close low-quality, high-churn customers just to hit the number, quietly damaging Churn Rate and long-term revenue even as the KPI looks great
Guarding against this usually means pairing a KPI with a “guardrail” metric — a second number that would catch the gaming behavior if the primary KPI were being optimized in isolation rather than in service of the real goal
Revisiting KPI definitions periodically, and asking “how could someone technically hit this number while making things worse,” is a useful habit for catching Goodhart’s Law risk before it does real damage

Why It Matters

KPIs turn vague goals like “grow the business” or “improve support” into something concrete enough to actually manage day to day
A well-chosen set of KPIs gives every function a clear, objective definition of what “doing well” means, reducing debates that would otherwise run on opinion and anecdote
Leading KPIs give teams enough lead time to correct course before a lagging outcome — a missed revenue target, a spike in churn — is already locked in
Distinguishing true KPIs from vanity metrics protects a company from mistaking activity or surface-level growth for actual progress
Consistent KPI tracking over time reveals trends a single snapshot can’t — whether a metric is improving, plateauing, or quietly deteriorating
Clear KPI ownership creates accountability: when a specific person or team owns a number, there’s always someone positioned to explain why it moved and what’s being done about it
Investors and boards read a company’s chosen KPIs as a signal of operational sophistication — tracking the right numbers, defined rigorously, suggests maturity beyond just headline revenue
KPIs make cross-functional tradeoffs visible: if a support KPI deteriorates while a sales KPI improves, leadership can see the tradeoff happening in real time instead of discovering it a quarter later

Common Pitfalls

Tracking vanity metrics as if they were KPIs: downloads, signups, or pageviews can dominate a dashboard simply because they’re big, flattering numbers, crowding out less impressive but far more meaningful figures
Choosing too many KPIs at once: a dashboard with thirty tracked numbers dilutes attention until nothing actually gets managed closely; a small, deliberately chosen set is more useful than a comprehensive one
Picking KPIs nobody can act on: a metric with no clear owner and no available lever to pull becomes background noise instead of a management tool
Only tracking lagging KPIs: a dashboard full of already-happened outcomes gives no early warning and leaves a team perpetually reacting instead of anticipating
Letting KPIs go stale: a metric that made sense at 10 employees may be meaningless at 200; KPIs should be revisited as the business and its goals evolve, not left on autopilot indefinitely
Confusing a KPI with the goal itself: a KPI is a proxy for progress, not the goal — optimizing the number in ways that technically move it while missing the underlying intent is a constant risk
Reporting KPIs inconsistently: changing how a KPI is calculated from quarter to quarter without flagging it makes trend lines meaningless and erodes trust in the reporting itself

KPI Dashboards and Review Cadence

Most functional teams review their core KPIs weekly, while company-wide KPIs tied to board reporting are typically reviewed monthly or quarterly
A good dashboard surfaces trend lines, not just the current snapshot — a KPI’s direction of travel is usually more informative than its absolute value at any single moment
Threshold-based alerting (flagging a KPI automatically when it crosses a set boundary) lets teams catch problems between scheduled reviews rather than waiting for the next meeting
Many teams color-code KPIs on a dashboard (on track, at risk, off track) so a reviewer can scan the overall picture in seconds before drilling into any single number that needs discussion
New employees onboard into a role faster when the KPIs that define success in that role are explicit from day one, rather than left as an unwritten, gradually absorbed expectation
Weekly or biweekly reviews work well for fast-moving, leading indicators, while lagging, outcome-level KPIs are often better suited to a monthly or quarterly cycle that matches how quickly they can realistically change
The best KPI reviews spend more time on “why did this move and what are we doing about it” than on simply reading the numbers aloud — the dashboard is a starting point for a conversation, not the conversation itself

Choosing Which KPIs to Track First

Early-stage teams should resist the urge to track everything measurable and instead pick the two or three KPIs most tied to the current biggest risk to the business — often activation or early retention before a product has proven it can keep users at all
As a company matures past its initial Product-Market Fit, the most important KPI often shifts from an activation or engagement metric toward an efficiency metric like CAC and LTV (Customer Acquisition Cost and Lifetime Value) or net revenue retention
A useful test when adding a new KPI to the dashboard: can the team name, right now, what action it would take if this number moved sharply in either direction? If not, it isn’t ready to be a tracked KPI yet
Removing a KPI that’s stopped being decision-relevant is just as important as adding new ones — a dashboard that only ever grows eventually becomes as unreadable as the eighteen-metric dashboard it was meant to replace

How KPI Focus Shifts by Company Stage

Pre-product-market fit: the most important KPIs are usually activation and early retention signals — is anyone actually using this, and do they come back — since growth and efficiency metrics are premature before the core product is sticky
Post-product-market fit, pre-scale: KPIs shift toward repeatable growth signals like ARR and MRR (Annual Recurring Revenue and Monthly Recurring Revenue) growth rate and CAC and LTV (Customer Acquisition Cost and Lifetime Value), since the goal becomes proving the growth engine works before pouring capital into it
Scale stage: KPIs increasingly emphasize efficiency and unit-level profitability — margins, payback periods, retention cohorts — since growth alone is no longer enough to satisfy investors or the board
Carrying pre-product-market-fit KPIs, like raw signups, forward into the scale stage as if they still mattered is a common way legacy vanity metrics linger long after they’ve stopped being useful

KPI Examples by Function

KPI is deliberately function-agnostic — unlike some of the more startup-specific metrics elsewhere in this vault, virtually every team in a company can define its own:

Sales: quota attainment, average deal size, sales cycle length, win rate against competitors
Product: activation rate (share of new users reaching a defined “aha moment”), feature adoption rate, time to first value
Support: first-response time, ticket resolution time, customer satisfaction score, ticket volume per active customer
Marketing: cost per lead, lead-to-customer conversion rate, organic traffic growth, email open and click-through rates, ROI (Return on Investment) by campaign
Engineering: deployment frequency, bug escape rate, system uptime, average time to resolve a production incident
Customer success: net revenue retention, expansion revenue per account, time to onboard, renewal rate ahead of contract expiration
Finance: Runway and Burn Rate, cash conversion cycle, gross margin
Growth and revenue: ARR and MRR (Annual Recurring Revenue and Monthly Recurring Revenue), Churn Rate, and CAC and LTV (Customer Acquisition Cost and Lifetime Value) are all KPIs in their own right, just specific and important enough to have earned their own dedicated terms

Example

A 40-person startup selling scheduling software walks into a quarterly planning meeting with a dashboard listing eighteen tracked numbers, including total app downloads, social media followers, total signups since launch, and total blog pageviews, alongside genuinely operational figures like MRR and churn. The new VP of Product proposes a cut: keep five KPIs per function on the main dashboard and move the rest to an optional secondary report nobody is required to review in the meeting itself.

For the product team, that means dropping “total signups” — a vanity metric that had been climbing steadily but wasn’t correlated with revenue — in favor of “trial-to-paid conversion rate” and “week-two activation rate,” two leading KPIs that predict revenue outcomes weeks before they show up in MRR. Within the first month of watching the new dashboard, the team notices week-two activation sitting at just 34%, far below their assumed 55%, even as trial signups (the old vanity metric) keep climbing every week. Digging in, they find a broken onboarding email that had been silently failing to send for six weeks — invisible on the old dashboard, glaring on the new one.

By the next quarterly review, week-two activation is up to 51% after fixing onboarding, and trial-to-paid conversion has followed it upward two weeks later, exactly as the leading-indicator relationship predicted. The lagging KPI, MRR growth, confirms the win a month after that. Attainment against the quarter’s activation target comes out to $\frac{51}{55} \times 100 \approx 92.7$ — short of target, but a dramatic improvement worth celebrating rather than a number to be discouraged by.

The VP uses the story in the next board meeting as the case for why the company now reviews five sharp KPIs per function instead of eighteen loosely related numbers: the smaller list is what actually caught and fixed a real problem, while the vanity metrics only ever told a comfortable story. One guardrail comes out of the postmortem too — support ticket volume tagged “onboarding confusion” is added as a secondary metric alongside activation rate, specifically so a future fix that boosts activation by quietly pressuring users through onboarding too fast would show up as a cost somewhere else on the dashboard, a small, deliberate defense against Goodhart’s Law.

Referenced by

Founders and Executives MOC

Dig deeper

Spot an error in this note?