KPI Design & Metric Frameworks
Topic 2 of 11 · In Data Driven Decisions · September 9, 2026
Design KPIs that actually drive behavior, use proven frameworks like OKRs and Balanced Scorecard, and avoid the vanity metrics that mislead everyone.
Overview
Key Performance Indicators (KPIs) are the compass of a data-driven organization. Without them, teams drift — busy but directionless. But poorly designed KPIs are worse than no KPIs at all: they reward the wrong behavior, create false confidence, and waste everyone's time chasing numbers that do not matter.
This topic teaches you to design KPIs that drive real outcomes. You will learn the difference between a metric and a KPI, how to write SMART KPIs, how to select the right framework (OKRs, Balanced Scorecard, North Star), and how to avoid the traps — vanity metrics, proxy metrics, and Goodhart's Law — that turn good intentions into perverse incentives. By the end, you will be able to build a KPI system that aligns strategy, execution, and measurement from the top floor to the front line.
Before You Start
- Have the revenue, churn and NPS figures from the previous topic's case studies ready to reference when you design KPIs.
- Be able to state the metric-versus-KPI test from memory: if the number doubled, would you change what you do?
- Practice the SMART acronym and score one vague goal ('improve customer experience') against all five letters before the lesson.
- Sketch the input → output → outcome pipeline and give one SaaS example of each stage.
- Write your company's strategy (or a invented mission) as a single sentence — you will cascade KPIs down from it.
Learning Objectives
- Distinguish between a raw metric and a true KPI
- Write KPIs that follow the SMART criteria
- Apply the Balanced Scorecard framework to measure business health across four perspectives
- Design OKRs (Objectives and Key Results) that cascade from company to team level
- Identify and select a North Star metric for a product or business
- Classify metrics as input, output, or outcome
- Detect and avoid vanity metrics, proxy metrics, and Goodhart's Law traps
- Build a cascading KPI hierarchy that connects strategy to daily work
Prerequisites
Start with Data-Driven Decision Making Fundamentals to understand decision types and maturity levels. Basic spreadsheet skills help but are not required.
Metric vs KPI: Not the Same Thing
A metric is any quantifiable measure: page views, email open rate, number of support tickets, temperature of a server. There are thousands of metrics in any business.
A KPI is a metric that has been deliberately chosen because it indicates whether the business is achieving a critical objective. Every KPI is a metric, but most metrics are not KPIs. Think of metrics as the full vocabulary and KPIs as the key sentences that carry meaning.
The test: if a metric goes up and you would change your behavior, it is a KPI. If you would not change anything, it is just a metric.
| Example | Metric or KPI? | Why |
|---|---|---|
| Total website visitors | Metric | Interesting, but high traffic does not equal success if none convert |
| Monthly recurring revenue (MRR) | KPI | Directly measures business health; you act on changes |
| Email open rate | Metric (borderline) | Useful for the marketing team but not a strategic indicator |
| Customer retention rate | KPI | Directly ties to revenue and lifetime value; drives decisions |
Writing SMART KPIs
Vague KPIs produce vague results. The SMART framework ensures every KPI is:
- Specific — clearly defines what is measured (not "improve sales" but "increase new customer revenue")
- Measurable — can be quantified with a number (not "better customer service" but "reduce average first-response time")
- Achievable — stretch but realistic given resources and constraints
- Relevant — directly tied to a business objective (not a vanity number)
- Time-bound — has a deadline or cadence ("by end of Q3", "month over month")
Bad KPI: "Get more customers." SMART KPI: "Increase monthly new customer sign-ups from 500 to 750 by end of Q2 through referral program optimization."
Notice the SMART version specifies what (sign-ups), from what baseline (500), to what target (750), by when (Q2), and how (referral program).
Metric Types: Input, Output, and Outcome
Every metric falls into one of three categories. Understanding the type tells you what action it supports:
| Type | Definition | Example | What it tells you |
|---|---|---|---|
| Input | Resources or efforts you invest | Hours of sales calls, ad spend, content published | Are we doing the work? |
| Output | Immediate results of the input | Leads generated, articles published, demos delivered | Is the work producing results? |
| Outcome | Business impact of the output | Revenue closed, customers retained, NPS score | Did the results matter? |
The ideal KPI system tracks outcomes as the primary signal, outputs as leading indicators, and inputs for accountability. Most teams get this backward — they measure inputs (activity) and assume outcomes follow.
Framework 1: Balanced Scorecard
The Balanced Scorecard, developed by Kaplan and Norton, measures business health across four perspectives to prevent over-focus on any single dimension:
- Financial — revenue growth, profitability, ROI. Are we financially healthy?
- Customer — satisfaction, retention, NPS, market share. Do customers value us?
- Internal Processes — efficiency, cycle time, defect rates. Are our operations effective?
- Learning & Growth — employee skills, innovation rate, culture. Can we sustain improvement?
Each perspective gets 2–3 KPIs. The framework forces balance: you cannot optimize profit (financial) while ignoring customer satisfaction or employee turnover. Most companies have strong financial KPIs and weak learning/growth KPIs — the Scorecard makes this gap visible.
Framework 2: OKRs (Objectives & Key Results)
OKRs are a goal-setting framework used by Google, Intel, Spotify, and thousands of companies. They work at every level:
- Objective — a qualitative, ambitious, inspiring goal ("Become the #1 choice for small business accounting")
- Key Results — 2–4 quantitative, measurable outcomes that prove you achieved the objective ("Increase SMB customer count from 2,000 to 5,000", "Achieve 90% onboarding completion rate", "Reduce churn from 8% to 3%")
Key OKR rules:
- OKRs are set quarterly, not annually — short cycles enable faster course correction
- A good OKR hit rate is 60–70% — if you hit 100%, your goals are too easy
- OKRs are separate from performance reviews — otherwise people sandbag
- Cascading OKRs: company → department → team → individual, each level supporting the one above
Framework 3: The North Star Metric
A North Star Metric (NSM) is a single metric that best captures the core value your product delivers to customers. It is not the only KPI — it is the one metric that, if it grows sustainably, indicates the entire business is healthy.
Examples:
- Airbnb: Nights booked (measures the core transaction that delivers value to hosts and guests)
- Spotify: Time spent listening (measures the core value delivery to users)
- Slack: Messages sent in teams (measures adoption and daily active usage)
- Facebook (Meta): Daily active users (measures whether people find ongoing value)
How to find yours: Ask: (1) What is the single action that means a customer got value? (2) If this number grows sustainably, does the business grow? (3) Can the team influence this metric with their work? If all three are yes, you have a candidate.
Anti-Patterns: Vanity, Proxy, and Goodhart's Law
Bad KPI design creates dysfunctional behavior. Watch for these traps:
Vanity metrics look impressive but do not drive decisions. "1 million page views" sounds great — but if none convert, it is meaningless. Test: if this number doubled, would you change what you do? If not, it is vanity.
Proxy metrics stand in for what you actually care about but can be gamed. Email open rate is a proxy for engagement — but subject line tricks ("You won't believe...") inflate opens while destroying trust.
Goodhart's Law: "When a measure becomes a target, it ceases to be a good measure." If you target "customer satisfaction scores," teams will game the survey. If you target "number of code deploys," developers will ship trivial changes. The defense: use multiple metrics, rotate targets, and pair quantitative KPIs with qualitative judgment.
Cascading KPIs: Strategy to Execution
A KPI cascade connects the CEO's vision to the front-line team's weekly work. The structure:
- Company KPIs — 3–5 metrics tied directly to the business strategy (e.g., Annual Recurring Revenue, Customer Lifetime Value, Net Promoter Score)
- Department KPIs — each department chooses 2–3 KPIs that directly influence a company KPI. Marketing: inbound lead volume. Product: feature adoption rate. Support: resolution time.
- Team/Individual KPIs — specific, operational metrics that each person can influence (e.g., content team: organic traffic per article; sales team: close rate per rep)
The cascade creates alignment: everyone can trace their daily work up to the company strategy. When a department hits its KPI but the company misses its target, the cascade reveals the broken link.
Hands-On Practice
Apply these frameworks to real or realistic business scenarios.
- Pick a company you admire. Find its North Star Metric (or make an educated guess). Explain why it captures core value delivery.
- Take a vague executive goal ("Improve customer experience") and write three SMART KPIs that would measure it.
- Build a mini Balanced Scorecard for a fictional SaaS startup: define one KPI for each of the four perspectives.
- Design OKRs for a marketing team supporting a company objective of "Double revenue in 12 months." Include at least one outcome-focused Key Result.
- Find a "vanity metric" in your own life or work. Explain why it is misleading and propose a better KPI to replace it.
Scenario: You are hired as the first data analyst at a growing e-commerce company. The CEO says: "We have no idea if we are doing well. Give me KPIs."
Your task:
- Interview the CEO and two department heads to understand the strategy
- Design 4 company-level KPIs using SMART criteria
- Create a cascading structure: company → marketing → operations → customer support
- Identify the North Star Metric
- For each KPI, name a counter-metric to prevent gaming
Deeper Worked Examples
Choosing a North Star metric for a plant-subscription retailer
Step 1
Step 2
Step 3
Step 4
Step 5
Step 6
Rebuilding a marketing OKR that was measuring output, not outcome
Step 1
Step 2
Step 3
Step 4
Step 5
Step 6
Common Mistakes
- Confusing metrics with KPIs — tracking 50 numbers is not the same as having 50 key indicators
- Writing KPIs that are not SMART — vague KPIs produce vague accountability
- Measuring only financial or outcome metrics and ignoring leading indicators (inputs/outputs)
- Setting KPIs without counter-metrics — enables gaming and perverse incentives
- Using KPIs for performance reviews — people will sandbag targets instead of stretching
- Copying another company's KPIs without adapting to your strategy and context
- Setting and forgetting — KPIs must be reviewed and updated as strategy evolves
Real-World Usage
Amazon's "Customer Obsession" principle is operationalized through a North Star metric: customer experience. Every team — from logistics to Prime Video — can trace their work to a customer-facing outcome. Their KPI cascade is legendary: the "flywheel" connects lower prices → more customers → more sellers → lower prices, and every team's KPIs feed into one or more loops. Starbucks uses a Balanced Scorecard approach, balancing financial metrics (same-store sales growth) with customer metrics (wait time, satisfaction) and employee metrics (retention, training hours). Google famously lets engineering teams set their own OKRs, but they must align upward — ensuring bottom-up innovation stays connected to top-down strategy.
Mini Project
Business problem: A SaaS startup selling project management tools has been growing for two years but the CEO says "I feel like we are flying blind." Revenue is growing but costs are growing faster, and nobody knows which metric actually predicts long-term success.
Objective: Design a complete KPI system using the frameworks from this topic.
Task 1
Task 2
Task 3
Task 4
Task 5
Task 6
| Deliverable | What it proves |
|---|---|
| North Star Metric | You can identify the single most important value metric |
| Balanced Scorecard | You can measure health across four perspectives |
| Team OKRs | You can cascade strategy to team-level goals |
| SMART targets | You can write precise, measurable KPIs |
| Counter-metrics | You can design KPIs that resist gaming |
| KPI playbook | You can package a system for organizational use |
Key Takeaways
- A metric is any number; a KPI is one the business actively tracks to judge success
- SMART criteria ensure KPIs are specific, measurable, achievable, relevant, and time-bound
- Input, output, and outcome metrics serve different roles — outcomes are primary, inputs are for accountability
- Balanced Scorecard balances financial, customer, process, and learning perspectives
- OKRs set ambitious quarterly goals with measurable key results that cascade from company to team
- A North Star Metric captures the single most important value your product delivers
- Vanity metrics, proxy metrics, and Goodhart's Law are the three enemies of good KPI design
- Cascading KPIs connect strategy to daily execution and reveal broken links
Key Terms & Glossary
| Term | Meaning |
|---|---|
| KPI vs metric | A metric is any quantifiable measure; a KPI is a metric deliberately chosen because a critical objective depends on it — the test is whether you would change behaviour if it moved. |
| SMART criteria | The requirement that a KPI be Specific, Measurable, Achievable, Relevant and Time-bound so it produces unambiguous accountability. |
| North Star metric | The single metric that best captures the core value a product delivers; if it grows sustainably the whole business is healthy. |
| Goodhart's Law | The principle that when a measure becomes a target it ceases to be a good measure, because people optimise the number rather than the goal it stood for. |
| Vanity metric | A number that looks impressive but would not change any decision if it doubled, such as raw page views with no conversion. |
| Counter-metric | A second KPI that detects when people have gamed the primary one, such as tracking survey response rates to stop cherry-picked customer scores. |
Learning Resources
How to Measure Anything — Douglas Hubbard
The definitive book on finding the value of anything, including intangibles like customer satisfaction and risk.
OKRs: The Simple Idea That Changed Google (What Matters)
John Doerr's explanation of the OKR framework and how it drove Google's growth from 40 to 40,000 employees.
Balanced Scorecard Institute
The official resource for the Balanced Scorecard framework, including templates and case studies.
North Star Metric — Product School
A practical guide to identifying and implementing your North Star Metric with real company examples.
Goodhart's Law (Wikipedia)
Background on the economic principle that explains why single-metric targets fail.
Chapter Quiz
Chapter Quiz — KPI Design & Metric Frameworks
Five questions on the difference between metrics and KPIs, SMART targets, metric types, North Stars, and Goodhart's Law.
Question 1 of 5
The test that separates a KPI from a plain metric is:
Knowledge Check
End-of-Topic Exam
Hint: Add the baseline, the mechanism, and a guardrail that would catch gaming of the primary number.
Model answer: A better statement: 'Grow new-business revenue from £1.6M to £2M (plus 25%) by the end of Q3 through expanding the mid-market team from four to six sellers.' A suitable counter-metric is average deal size or gross margin on closed deals, because a team chasing pure revenue could sign low-quality, discounted deals to hit the target. The counter-metric ensures the target is reached profitably, not just nominally.
Hint: Think of tactics that inflate opens without engaging the reader.
Model answer: Open rate is a proxy for engagement: it stands in for the real goal but can be manipulated. A team could write deceptive subject lines ('you won't believe this'), mislabel emails as conversations, or split sends across the day to catch inboxes, all of which raise opens while damaging trust. It should be replaced by an outcome measure such as replies, click-throughs to a product page, or net promoter-style engagement that links directly to behaviour and revenue.
Hint: A cascade that is internally consistent but misaligned to the strategy gives this exact symptom.
Model answer: The most likely cause is a broken link in the cascade: department KPIs were chosen to be achievable in isolation rather than to influence company revenue, so each team optimises its own line while no-one owns the connecting outcome. The diagnostic step is to trace each departmental KPI upward to a company KPI and ask whether hitting it plausibly moves revenue; any department whose KPI cannot be mapped is the broken link and its KPI should be redesigned.