Growth Metrics, Analytics and Unit Economics

Introduction

Growth is not synonymous with rapid acquisition. A venture grows sustainably when it creates customer value, converts that value into repeatable revenue, and does so with an economically viable use of capital. Growth metrics provide evidence about these processes; analytics explains the relationships among them; and unit economics tests whether expansion improves or destroys value.

For an entrepreneurial manager, the central question is not “Which number is increasing?” but “Which customer and economic mechanism is producing the increase, and will it remain attractive as the venture scales?” This chapter develops a measurement system built around a North Star Metric, metric trees, funnels, cohorts, attribution, customer acquisition cost, lifetime value, churn, payback, contribution margin, and burn efficiency.

Learning Objectives

By the end of this chapter, learners should be able to:

1. Define an appropriate North Star Metric and construct a metric tree.
2. Analyse acquisition, activation, conversion, retention, referral and revenue funnels.
3. Use cohort analysis to distinguish sustainable retention from temporary growth.
4. Evaluate attribution methods and recognise their limitations.
5. Calculate and interpret CAC, LTV, churn, payback period and contribution margin.
6. Assess burn efficiency and use unit economics in scaling decisions.
7. Design a responsible measurement system that addresses data quality, privacy and experimentation risks.

1. From Activity Measures to Value Measures

A metric is a quantified indicator used to monitor a phenomenon or support a decision. Metrics are useful only when their definitions, time periods, populations and decision rules are explicit. “Users,” for example, may mean registered accounts, active accounts, paying customers or unique people. These are not interchangeable.

1.1 Vanity and actionable metrics

Vanity metrics—such as total downloads, registered users or social impressions—may describe scale but often do not explain value creation. Actionable metrics are sufficiently specific and timely to guide intervention. Examples include activation rate, week-four retention, gross-margin-adjusted CAC, or the percentage of orders delivered within the promised time.

A metric should normally have:

• a precise numerator and denominator;
• a defined unit and time window;
• a specified population and inclusion rule;
• a reliable data source;
• an owner and reporting frequency; and
• a decision threshold or intended use.

Metrics must also be classified. Input metrics measure controllable activities, such as qualified sales conversations. Output metrics measure results, such as new customers. Outcome metrics measure durable value, such as retained contribution profit. A team that optimises outputs without observing outcomes may acquire customers who never produce economic value.

2. The North Star Metric and Metric Tree

2.1 North Star Metric

The North Star Metric (NSM) is the principal measure of customer value delivered by the venture over a defined period. It should reflect the product’s core value exchange rather than merely internal activity or short-term revenue.

Examples include:

• completed and successful learning sessions for an education platform;
• weekly active teams completing a meaningful workflow for collaboration software;
• successfully delivered orders for a marketplace; and
• nights stayed with satisfactory service for a lodging platform.

A strong NSM has four properties: it represents value to the customer, is connected to the business model, is sensitive to product improvement, and is difficult to inflate without delivering real value. It should be complemented by guardrail metrics such as refund rate, complaint rate, gross margin, safety incidents and service quality.

The NSM is not necessarily the same as revenue. Revenue may be an important business outcome, but it can increase through price changes, one-time purchases or aggressive discounting while customer value declines.

2.2 Constructing a metric tree

A metric tree decomposes a high-level objective into drivers that teams can influence. For example:

Figure: A simplified metric tree linking customer behaviour to economic value.

The tree must be mathematically and conceptually coherent. If a business objective is revenue, a basic decomposition is:

This identity is useful only if the population and period are consistent. For a subscription business, revenue may instead be modelled through active subscribers, subscription price, expansion revenue and credits. A metric tree should expose assumptions, not disguise them.

3. Funnels and Conversion Analysis

A funnel represents sequential stages through which a prospect or customer passes. A typical digital funnel is awareness → visit → sign-up → activation → first transaction → repeat use → referral. A B2B funnel may be lead → qualified lead → opportunity → proposal → closed-won account → renewal.

For a stage conversion rate:

The numerator must be a subset of the denominator, and both must refer to the same cohort or observation period. If 2,000 visitors produce 160 sign-ups, visitor-to-sign-up conversion is 160 ÷ 2,000 = 8%. If 40 of those sign-ups activate, sign-up-to-activation is 40 ÷ 160 = 25%. The combined visitor-to-activation conversion is 40 ÷ 2,000 = 2%, not 8% + 25%.

Funnels support diagnosis. A large fall between sign-up and activation may indicate poor onboarding, unclear value or technical friction. However, funnel analysis can mislead when users skip stages, return later, appear on multiple devices, or are counted in different time windows. Event definitions and identity resolution are therefore essential.

Managers should examine both conversion and volume. Improving a low-volume channel from 20% to 40% may matter less than improving a high-volume channel from 5% to 7%. Statistical uncertainty should also be considered, especially with small samples.

4. Cohort Analysis and Retention

A cohort is a group sharing a meaningful starting event, usually acquisition month, first purchase month or activation week. Cohort analysis compares behaviour over equivalent customer ages. It prevents a growing business from confusing new-customer volume with improving customer quality.

For a retention cohort:

If 1,000 customers acquired in January contain 420 active customers in week four, week-four retention is 42%. The cohort’s denominator remains the original 1,000; it is not replaced by the current total user base.

Retention may be measured by activity, revenue, subscription status or repeated value delivery. The chosen definition must reflect the product. For a food-delivery service, monthly ordering may be appropriate; for emergency software, usage frequency may be low but contract renewal may be the meaningful outcome.

A retention curve can show:

• early drop-off, often associated with weak activation or poor expectation setting;
• stabilisation, suggesting a retained core; and
• continued decay, which may make long-term LTV assumptions unreliable.

Gross revenue retention excludes expansion from existing customers and is especially useful for subscription health. Net revenue retention includes expansion, downgrades and contraction, and may exceed 100% even when some customers churn. These measures should not be confused with logo retention, which counts customers rather than revenue.

5. Attribution: Linking Outcomes to Causes

Attribution assigns credit for a conversion or other outcome to marketing, sales or product touchpoints. It is a decision model, not a direct observation of causality.

Common models include:

A platform may report “return on ad spend” using its own attribution window, while the finance team uses cash receipts and incremental contribution margin. These figures answer different questions. Incrementality asks what would not have happened without the intervention and is better assessed through experiments, geo-tests, holdouts or carefully designed quasi-experiments.

Attribution is particularly fragile when cookies are blocked, devices are shared, channels interact, offline sales occur, or retargeting reaches customers who would have purchased anyway. Managers should avoid treating attributed revenue as causal revenue.

6. Unit Economics

Unit economics evaluates the revenue and variable cost associated with one economic unit: a customer, order, subscription, transaction or account. The unit must be chosen carefully. A marketplace may need separate economics for the buyer, seller and transaction.

6.1 Customer acquisition cost

A practical blended CAC is:

The numerator should specify whether it includes media, agency fees, sales salaries, commissions, onboarding and promotional incentives. The denominator must use the same period and customer definition. Blended CAC is useful for total planning; channel CAC supports allocation decisions. CAC calculated from attributed conversions is not necessarily incremental CAC.

6.2 Contribution margin

Contribution margin is the amount remaining after variable costs that vary with the unit sold or served:

Variable costs may include payment processing, fulfilment, cloud consumption, usage-based support, refunds and partner commissions. The classification depends on the decision horizon. A salaried employee may be fixed in the short run but avoidable in a longer capacity decision.

Example: A subscription produces monthly revenue of ₹1,000. Payment fees are ₹30, usage costs ₹120, and variable support costs ₹50. Contribution margin is ₹1,000 − ₹200 = ₹800, and the contribution margin ratio is 80%.

6.3 Churn

For customer churn over a period:

This simple formula assumes the denominator excludes newly acquired customers and that “lost” is clearly defined. If a business has significant additions during the month, ending customers cannot be used as a substitute for beginning customers without creating distortion.

For a stable subscription base with constant periodic churn c, an approximate expected lifetime in periods is:

This is a simplifying approximation based on a constant hazard, independent periods and no expansion, contraction or reactivation. It should not be applied blindly to heavily seasonal or contract-based businesses.

6.4 Lifetime value

A contribution-based LTV estimate for a simple subscription business is:

Under the constant-churn approximation:

If monthly revenue is ₹1,000, contribution margin ratio is 80%, and monthly churn is 5%, then:

• expected lifetime ≈ 1 ÷ 0.05 = 20 months;
• LTV ≈ ₹1,000 × 0.80 × 20 = ₹16,000.

This is an expected contribution value before acquisition cost, fixed overhead and financing cost. It assumes constant price, margin and churn, and does not model discounting. For a more formal discounted model, if contribution in period t is C_t and discount rate per period is r:

The finite horizon T is selected according to the decision context. Forecast uncertainty should be represented with sensitivity analysis rather than a single falsely precise number.

6.5 LTV:CAC and payback

The LTV:CAC ratio is:

It is meaningful only when LTV and CAC use comparable customer populations, margins, time periods and attribution rules. A high ratio may reflect underinvestment in growth, while a low ratio may reflect immature retention rather than a permanently bad model.

Payback period is the time required for cumulative contribution from a newly acquired customer to recover CAC. If monthly contribution is constant m and CAC is K:

With CAC of ₹6,000 and monthly contribution of ₹800, payback is 7.5 months. If contribution varies by month, cumulative contribution should be calculated month by month; the payback point is the first period in which cumulative contribution equals or exceeds CAC. Cash timing, annual prepayment and refunds can materially change cash payback.

7. Burn Efficiency and Scaling Decisions

Burn is the net cash consumed during a period. A simple net burn calculation is:

A commonly used growth efficiency indicator is the burn multiple:

If a venture burns ₹12 million in a quarter and adds ₹6 million of net new ARR, its burn multiple is 2.0. The measure is meaningful only when ARR is correctly defined, revenue quality is sound, and the same period is used. It can be distorted by annual contracts, seasonality, acquisition timing, working-capital movements and revenue recognition.

Managers should read burn efficiency alongside gross margin, retention, payback and runway. Runway in months is approximately:

This assumes burn is stable. A forecast should instead model hiring, inventory, seasonality, fundraising timing and downside scenarios.

8. Integrated Managerial Use

Metrics become valuable when connected to decisions. A practical operating sequence is:

1. Define the customer value event and NSM.
2. Build a metric tree showing behavioural and economic drivers.
3. Specify event definitions, ownership and data lineage.
4. Review acquisition and activation funnels.
5. Compare retention and revenue cohorts.
6. Estimate incremental CAC and contribution-based LTV.
7. Monitor payback, burn and runway.
8. Run experiments against a pre-stated decision metric and guardrails.
9. Reforecast when observed cohorts differ from assumptions.

A growth team might test a shorter onboarding process. The primary metric could be activated users within seven days; guardrails could include fraud, refund, support-contact and week-four retention rates. A statistically significant improvement in activation that reduces retention or increases costly support is not necessarily a successful intervention.

9. Illustrative Case: LearnLoop

Background

LearnLoop is a fictional subscription platform offering short professional courses. It charges ₹1,200 per month. Payment fees, content delivery and variable support total ₹300 per subscriber per month, giving a contribution of ₹900 and a contribution margin ratio of 75%.

Problem

The venture has grown rapidly through paid advertising but is uncertain whether it should increase its monthly advertising budget. The management team reports a 3:1 LTV:CAC ratio using revenue rather than contribution margin.

Analysis

In the latest month, advertising and commissions total ₹1.8 million and produce 600 new paying subscribers. Blended CAC is therefore:

The observed monthly churn of the relevant mature cohort is 6%. Using the constant-churn approximation:

The contribution-based LTV:CAC ratio is 5:1, not 3:1. However, cohort data show that new customers churn at 10% in the first three months before stabilising. A single mature-cohort churn rate may therefore overstate LTV for newly acquired customers.

Suppose contribution is ₹900 in month one, ₹810 in month two, ₹729 in month three, and then ₹900 per month thereafter if retained. Payback occurs when cumulative contribution exceeds ₹3,000: after month one cumulative contribution is ₹900, after month two ₹1,710, after month three ₹2,439, and after month four ₹3,339. Payback is therefore reached during month four, subject to the stated retention and contribution assumptions.

Approach and decision

LearnLoop should not scale solely on the headline ratio. It should:

• calculate CAC separately for each channel using incremental tests where feasible;
• use acquisition cohorts rather than mature customers for initial LTV;
• track activation as completion of a meaningful lesson and link it to retention;
• test onboarding changes with week-four retention as a guardrail;
• model cash payback, including annual discounts and payment timing; and
• increase spend gradually only in channels whose marginal cohorts maintain acceptable payback and retention.

Learning points

The case illustrates four principles. First, LTV must be contribution-based when used to evaluate acquisition. Second, averages conceal cohort variation. Third, payback is a cash and timing measure, not merely a profitability ratio. Fourth, growth investment should be scaled using marginal evidence rather than historical blended averages.

10. Data Governance, Ethics and Contemporary Practice

Analytics systems can create privacy, fairness and manipulation risks. Teams should collect only data necessary for a legitimate purpose, disclose material tracking, apply access controls, retain data no longer than needed, and respect applicable privacy and consumer-protection law. Personally identifiable data should be separated from analytical identifiers where possible.

Metric design can produce harmful incentives. A delivery company that optimises orders per courier without a safety guardrail may encourage unsafe behaviour. A lending platform that optimises approval volume without monitoring repayment fairness may discriminate or increase harm. North Star Metrics should therefore be accompanied by quality, safety, inclusion and compliance guardrails.

AI can assist with anomaly detection, forecasting, segmentation and natural-language querying, but generated conclusions require validation. Data drift, biased historical labels, leakage from future information and spurious correlations can produce confident but defective decisions. Human accountability remains necessary for consequential decisions.

Contemporary measurement practice is moving toward privacy-preserving experimentation, server-side event collection, incrementality testing, causal inference, real-time product analytics and finance-grade metric definitions. These tools improve evidence but do not remove the need for sound hypotheses and managerial judgement.

11. Limitations and Best Practices

Unit economics is a model, not a physical law. It is sensitive to customer heterogeneity, price changes, competition, capacity constraints, seasonality and accounting definitions. LTV estimates are especially unstable when cohorts are young. Attribution can encourage channel conflict and double counting. Funnel percentages can conceal absolute volume and sampling bias.

Best practices include:

• maintain a metric dictionary with formulas, owners and data sources;
• reconcile product, marketing and finance numbers regularly;
• segment by channel, geography, plan, customer type and acquisition cohort;
• report distributions and confidence intervals where appropriate, not only averages;
• distinguish descriptive attribution from causal incrementality;
• use contribution margin rather than revenue for acquisition decisions;
• forecast base, upside and downside cases;
• investigate metric discontinuities and instrumentation changes; and
• stop or redesign experiments that harm safety, privacy or customer welfare.

Chapter Summary

A sustainable growth system measures customer value, not activity alone. The North Star Metric expresses the venture’s central value delivery, while a metric tree decomposes it into controllable behavioural and economic drivers. Funnels reveal stage-level conversion, and cohort analysis reveals whether retention and revenue improve for comparable customer groups. Attribution helps organise touchpoints but does not prove causality; incrementality requires experiments or stronger causal methods.

Unit economics connects growth to financial viability. CAC measures acquisition investment per new customer, contribution margin measures value remaining after variable cost, churn describes customer loss, and contribution-based LTV estimates the future value of a customer under explicit assumptions. Payback assesses the timing of recovery of acquisition investment, while burn efficiency relates cash consumption to durable revenue growth. These metrics must be interpreted together, with consistent definitions, sensitivity analysis, guardrails and responsible data governance.

Key Terms

North Star Metric: Principal measure of customer value delivered.

Metric tree: Hierarchical decomposition of an objective into measurable drivers.

Actionable metric: Measure that can guide a specific managerial intervention.

Funnel: Sequential stages from initial contact to a desired outcome.

Activation: Achievement of an early event indicating meaningful product value.

Cohort: Group sharing a defined starting event or characteristic.

Retention: Continued activity, subscription or value relationship over time.

Churn: Loss of customers or recurring revenue during a period.

Attribution: Assignment of credit for an outcome to one or more touchpoints.

Incrementality: Outcome caused by an intervention beyond what would otherwise occur.

CAC: Acquisition and sales cost per newly acquired customer.

LTV: Expected future value generated by a customer, commonly measured as contribution.

Contribution margin: Revenue remaining after variable costs.

Payback period: Time required for cumulative contribution to recover CAC.

Burn: Net cash consumed by a venture during a period.

Burn multiple: Net burn divided by net new recurring revenue.

Guardrail metric: Measure used to detect harmful or unacceptable side effects.

Metric dictionary: Controlled record of metric definitions, sources and ownership.

Review Questions

Short Answer Questions

1. What distinguishes a North Star Metric from a vanity metric?
2. Why must retention be calculated against the original cohort size?
3. Define contribution margin and explain why it is used in LTV.
4. What is the principal limitation of last-touch attribution?
5. State two assumptions behind the approximation Expected lifetime ≈ 1 ÷ churn.

Long Answer Questions

1. Explain how to construct a metric tree for a subscription business, including behavioural, economic and guardrail metrics.
2. Compare funnel analysis and cohort analysis, and explain why a growth manager needs both.
3. Discuss attribution models and explain why attributed revenue should not automatically be interpreted as incremental revenue.
4. Derive a contribution-based LTV estimate for a subscription business and critically assess its limitations.
5. Explain how CAC, payback, burn multiple and runway should jointly inform a scaling decision.

Analytical / Application Questions

1. A service charges ₹800 per month, has a contribution margin ratio of 70%, monthly churn of 4%, and CAC of ₹8,000. Calculate approximate LTV, LTV:CAC and payback assuming constant monthly contribution. State whether the result justifies immediate scaling and identify two additional data requirements.
2. A funnel has 50,000 visits, 4,000 registrations, 1,000 activations and 250 paid customers. Calculate each sequential conversion rate and the overall visit-to-paid conversion rate. Identify the stage with the largest absolute loss and explain why that may not be the best improvement target.
3. Two channels each acquire 100 customers. Channel A has CAC of ₹2,000 and month-six retention of 30%; Channel B has CAC of ₹3,000 and month-six retention of 55%. Design a decision framework that avoids selecting a channel using CAC alone, specifying the additional unit-economic and causal evidence required.

Multiple Choice Questions with Answers

Question 1

Which statement best describes a North Star Metric?

A. It is the largest number displayed on a dashboard.

B. It measures customer value delivered and connects to the business model.

C. It is always identical to total revenue.

D. It measures only employee activity.

Correct Answer: B. It measures customer value delivered and connects to the business model.

Question 2

A funnel receives 2,000 visitors and produces 160 sign-ups. What is the visitor-to-sign-up conversion rate?

A. 4%

B. 8%

C. 12.5%

D. 16%

Correct Answer: B. 8%

Question 3

A January cohort contains 1,000 customers, of whom 420 are active in week four. What is week-four retention?

A. 4.2%

B. 42%

C. 58%

D. 70%

Correct Answer: B. 42%

Question 4

Which attribution method gives all conversion credit to the final recorded touchpoint?

A. First-touch attribution

B. Linear attribution

C. Last-touch attribution

D. Time-decay attribution

Correct Answer: C. Last-touch attribution

Question 5

A venture spends ₹900,000 on acquisition and gains 300 new customers. What is CAC?

A. ₹300

B. ₹1,500

C. ₹3,000

D. ₹270,000

Correct Answer: C. ₹3,000

Question 6

A subscription earns ₹1,000 monthly revenue and has variable cost of ₹250. What is its contribution margin ratio?

A. 25%

B. 50%

C. 75%

D. 80%

Correct Answer: C. 75%

Question 7

If monthly churn is 5%, what is the approximate expected lifetime under the constant-churn approximation?

A. 5 months

B. 10 months

C. 20 months

D. 95 months

Correct Answer: C. 20 months

Question 8

A customer has CAC of ₹4,000 and generates constant monthly contribution of ₹1,000. What is simple payback?

A. 2 months

B. 4 months

C. 5 months

D. ₹4,000 months

Correct Answer: B. 4 months

Question 9

Which measure most directly tests whether a marketing intervention caused additional conversions?

A. Last-touch attribution

B. Total impressions

C. Incrementality testing

D. Registered-user count

Correct Answer: C. Incrementality testing

Question 10

A company burns ₹10 million in a quarter and adds ₹5 million of net new ARR. What is its burn multiple?

A. 0.5

B. 2.0

C. 5.0

D. ₹15 million

Correct Answer: B. 2.0