Entrepreneurship
Lean Analytics
by Alistair Croll and Benjamin Yoskovitz
Lean Analytics summary
Lean Analytics by Alistair Croll and Benjamin Yoskovitz is a practical framework for using data to reduce the biggest uncertainty in a startup or new product. Its central argument is not “measure everything.” It is the opposite: understand your business model, identify the stage you are actually in, and focus the organization on the One Metric That Matters (OMTM) for the question that matters now.
The book combines Lean Startup thinking with analytics. Early teams often collect dozens of dashboards while still being unclear about the problem they are solving, whether users return, why growth is happening, or whether revenue can become sustainable. Lean Analytics gives those questions an order. The authors describe five broad stages—Empathy, Stickiness, Virality, Revenue, and Scale—and argue that each stage deserves a different kind of evidence.
This is important because a metric can be useful in one stage and distracting in another. Customer lifetime value is not the first question when you are still unsure whether anyone has a painful problem. Viral growth is dangerous if the product cannot retain users. Revenue growth can hide bad unit economics. The discipline is therefore to ask: What is the riskiest assumption right now, and what measurement would tell us whether it is becoming less risky?
Lean Analytics is best read as a decision system rather than a catalog of KPIs. It teaches founders to distinguish useful metrics from vanity metrics, compare cohorts instead of relying only on totals, set explicit targets or “lines in the sand,” and change focus as evidence moves the company to a new stage.
Detailed summary
Lean Analytics — Premium Deep Summary
Overview
Lean Analytics by Alistair Croll and Benjamin Yoskovitz is a practical framework for using data to reduce the biggest uncertainty in a startup or new product. Its central argument is not “measure everything.” It is the opposite: understand your business model, identify the stage you are actually in, and focus the organization on the One Metric That Matters (OMTM) for the question that matters now.
The book combines Lean Startup thinking with analytics. Early teams often collect dozens of dashboards while still being unclear about the problem they are solving, whether users return, why growth is happening, or whether revenue can become sustainable. Lean Analytics gives those questions an order. The authors describe five broad stages—Empathy, Stickiness, Virality, Revenue, and Scale—and argue that each stage deserves a different kind of evidence.
This is important because a metric can be useful in one stage and distracting in another. Customer lifetime value is not the first question when you are still unsure whether anyone has a painful problem. Viral growth is dangerous if the product cannot retain users. Revenue growth can hide bad unit economics. The discipline is therefore to ask: What is the riskiest assumption right now, and what measurement would tell us whether it is becoming less risky?
Lean Analytics is best read as a decision system rather than a catalog of KPIs. It teaches founders to distinguish useful metrics from vanity metrics, compare cohorts instead of relying only on totals, set explicit targets or “lines in the sand,” and change focus as evidence moves the company to a new stage.
Critical synthesis
The real problem Lean Analytics solves
Most organizations do not suffer from a lack of data. They suffer from a lack of decision focus. A startup can instrument every click and still avoid the fundamental question of whether it is solving a real problem. A product team can celebrate rising monthly users while retention is weakening. A marketing team can reduce acquisition cost while attracting customers whose lifetime value is even lower.
Lean Analytics tries to stop this form of self-deception. The book’s deepest contribution is the connection between business model + stage + metric. A metric has meaning only in relation to the question the company is trying to answer.
1. One Metric That Matters is a focusing device
The OMTM is often misunderstood as “a business needs only one KPI.” That is not the argument. Companies should retain diagnostic data and monitor important health metrics. The OMTM is the number that receives the most attention because it best captures the current constraint.
For an early product, that might be the percentage of interviewed customers who experience a particular problem with sufficient intensity. Later it might be weekly retention, invitation rate, conversion to paid, contribution margin, or another metric tied to the stage. The point is to create organizational alignment: if everyone knows the current question and the metric that answers it, teams are less likely to optimize unrelated numbers.
OMTM also creates a natural review cycle. Once the metric crosses a meaningful threshold, the company asks whether the dominant risk has changed. Focus moves; it is not permanently attached to one dashboard.
2. Stage One — Empathy: prove the problem deserves to exist
In the Empathy stage the team is learning about customers, not proving that its solution is brilliant. Qualitative evidence is especially valuable because the unknowns are still broad. Interviews, observation, problem discovery, and early solution tests help determine whether the proposed problem is real, frequent, painful, and shared by a sufficiently useful market.
This stage is where many “data-driven” teams make a category error. They want statistically clean product metrics before they have a product or stable behavior to measure. Lean Analytics instead treats conversations and structured qualitative evidence as legitimate data. The important discipline is to avoid leading questions and to look for behavior rather than polite enthusiasm.
3. Stage Two — Stickiness: prove that the product creates repeat value
Once a product exists, the key question becomes whether users actually derive enough value to return. This is where retention and engagement matter. Signups alone can be a vanity metric because they measure entry, not sustained value.
Cohort analysis is especially powerful here. Suppose total weekly active users increase for four months. That looks healthy. But if each new cohort retains worse than the previous one and growth is maintained only by spending more on acquisition, the product may be deteriorating. Cohorts make the hidden pattern visible.
Stickiness is also contextual. A payroll product, a daily social app, and an annual tax tool should not have the same expected frequency. The correct metric reflects the natural usage cycle of the product.
4. Stage Three — Virality: understand whether users create more users
Virality can dramatically reduce acquisition friction, but the authors place it after stickiness for a reason. If a product has poor retention, stronger referral loops simply send more people into a leaky system.
Useful viral measures include how many invitations users send, the fraction accepted, the viral coefficient, and the time required for a viral cycle. But the number should connect to a real mechanism. A one-time referral promotion is not the same as a product whose value naturally increases when more participants join.
The broader lesson is that growth metrics should explain why growth happens. A rising chart without a mechanism is fragile because the team does not know which behavior to preserve.
5. Stage Four — Revenue: prove a repeatable business, not just a product
At the Revenue stage the company needs evidence that it can convert product value into an economically sustainable model. Gross revenue is useful, but it can still mislead. If a company spends more to acquire customers than those customers generate, growth can destroy value.
Customer acquisition cost (CAC), customer lifetime value (LTV), gross margin, conversion rate, churn, revenue per customer, and payback time are examples of metrics that help explain economics. Which one becomes the OMTM depends on the model and the current bottleneck.
A particularly useful insight is that “making money” and “having a scalable business” are different. A business may have customers but still depend on founder effort, unusually cheap acquisition, unsustainable discounts, or long payback periods. Revenue-stage analytics should expose those dependencies.
6. Stage Five — Scale: optimize a model that has already earned the right to grow
Scale is not merely “more traffic.” It means the company has enough evidence about product value and economics to widen the system. The challenge becomes finding additional channels, markets, segments, operational capacity, and efficiency without breaking what worked at smaller size.
At this point a single startup-wide metric may become less sufficient because the organization has multiple mature functions. Still, the OMTM principle remains useful inside teams: identify the current constraint rather than allowing every department to optimize a local number that may conflict with the whole business.
7. Business model changes the meaning of “good”
Lean Analytics compares several business-model patterns because the same metric does not carry the same meaning everywhere. A SaaS product cares deeply about recurring revenue, churn, expansion, and acquisition economics. An e-commerce business must understand conversion, repeat purchase, average order value, and margins. A marketplace has to balance supply and demand, liquidity, transaction frequency, and take rate. Media and user-generated-content businesses have different engagement and monetization structures.
This is why generic benchmark lists can be dangerous. A “good” conversion rate is not meaningful without knowing the model, price point, acquisition channel, customer intent, and stage.
8. Vanity metrics versus actionable metrics
Vanity metrics are not necessarily false. They are often incomplete. Total registered users, cumulative downloads, pageviews, or followers can legitimately increase while the underlying business weakens. Their main problem is that they are easy to celebrate and hard to act on.
An actionable metric is connected to a decision. If weekly retention falls from one cohort to the next, a product team can investigate onboarding or product value. If trial-to-paid conversion differs sharply by acquisition channel, marketing and product teams can change channel mix or positioning. If CAC payback becomes too long, the business can adjust acquisition, price, margin, or retention.
9. Leading and lagging indicators
Lagging indicators tell you what has already happened—revenue, churn, or closed sales, for example. Leading indicators may provide earlier evidence about what is likely to happen next, such as activation behavior that historically predicts retention.
The trap is to treat correlation as certainty. A leading indicator is useful only while its relationship with the desired outcome remains reliable. Teams should periodically test whether the proxy still predicts what they care about.
10. Why cohorts beat aggregates
Aggregates mix users with different histories. A product may appear to have 40% retention because long-time loyal users are combined with new users who are leaving rapidly. Cohorts group people by a shared start time or characteristic, making change easier to see.
This is one of the book’s most transferable practices. Cohort thinking applies beyond startups—to email engagement, education, subscriptions, marketplaces, communities, and even content platforms. Whenever a total number combines populations exposed to different versions or conditions, cohort analysis can reveal whether the system is actually improving.
11. Draw a line in the sand
Before running an experiment, decide what result would be strong enough to justify the next step. This protects the team from changing the interpretation after seeing the outcome.
The threshold does not have to be perfect. It can come from industry benchmarks, comparable products, historical baselines, unit economics, or explicit management judgment. Its value is epistemic discipline: the team knows in advance what it is trying to learn.
If the result misses the line, the right response is not automatically to abandon the product. It is to investigate why: wrong segment, weak value proposition, poor execution, inappropriate threshold, or false assumption. The result creates a better question.
12. Data-driven versus data-informed
A mature reading of Lean Analytics is not “numbers should make every decision.” Data can describe behavior but may not capture strategy, ethics, brand, long-term bets, or phenomena for which the sample is too small. The practical goal is to become data-informed: use measurement to challenge intuition without pretending that measurement eliminates judgment.
Founders still choose which market to pursue, what kind of company to build, and which trade-offs are acceptable. Analytics improves those decisions by making assumptions visible.
Where the framework is strongest
Lean Analytics is especially useful when a team is overwhelmed by dashboards or is arguing about priorities. The stage framework gives the conversation order. OMTM makes the bottleneck explicit. Cohorts and comparative metrics reduce the seduction of cumulative growth charts. Lines in the sand reduce post-hoc rationalization.
Where to be cautious
The five stages are a model, not a law of nature. Real companies may revisit earlier stages, run multiple products, monetize before virality, or operate in markets where regulation, enterprise sales cycles, seasonality, or network effects change the sequence. A mature organization can have different products in different stages at the same time.
Benchmarks can also become false precision. What matters is whether the chosen metric links to the economics and behavior of the specific product. A team should use the framework to improve questions, not to force reality into a template.
A practical synthesis
The book can be compressed into a recurring loop:
- Identify the riskiest business assumption.
- Determine the stage and business model that shape the question.
- Select the metric that best represents progress on that risk.
- Establish a baseline and a target before acting.
- Run a focused experiment.
- Compare cohorts or segments where aggregate data could mislead.
- Decide whether to keep, change, or reject the hypothesis.
- When the bottleneck changes, change the OMTM.
That loop is more valuable than memorizing a long list of startup metrics. It turns analytics from reporting into a system for learning.
Key ideas
- One Metric That Matters (OMTM): A company will track many numbers, but at a given stage one metric should receive disproportionate attention because it best represents the current bottleneck or risk. OMTM creates focus; it is meant to change as the company progresses.
- Know your stage before choosing your metric: The five Lean Analytics stages are Empathy, Stickiness, Virality, Revenue, and Scale. Measuring the wrong stage can lead to premature optimization—for example, buying acquisition before proving retention.
- Empathy comes before growth: Early work is largely qualitative. The team needs to understand whether a real group of people has a meaningful problem and whether the proposed solution addresses it. Interviews and direct observation matter before sophisticated dashboards.
- Stickiness means proving that people get value repeatedly: Retention, engagement, frequency, activation, and cohort behavior become more important than raw signups. If users do not return, accelerating acquisition usually magnifies the leak.
- Virality is a growth mechanism, not a substitute for value: Referral loops and viral coefficients matter only after the product is useful enough to retain people. Growth that brings users into a weak product can create impressive top-line charts without a durable business.
- Revenue must be economically healthy: At the Revenue stage the company has to prove that monetization can work consistently. Metrics such as customer acquisition cost, customer lifetime value, margins, conversion, and time to recover acquisition spend become central.
- Scale changes the optimization problem: After a business has a repeatable model, the questions shift toward efficiency, predictability, market expansion, channels, operations, and sustainable growth. A metric that was useful for discovery may no longer be the right operating metric.
- Good metrics are comparative and actionable: Ratios, rates, segments, and cohorts often reveal more than cumulative totals. A useful metric should help a team decide what to do differently rather than simply report that activity occurred.
- Cohort analysis separates product improvement from accumulated growth: Comparing users who started in different periods can show whether retention or behavior is actually improving. A rising total user count can hide deteriorating cohort quality.
- Set a line in the sand before interpreting results: Define the threshold that would count as good enough before the experiment ends. Otherwise teams can rationalize almost any result after seeing the data.
Practical application
Apply Lean Analytics to a real project
1. Name the stage. Decide whether your biggest uncertainty is problem discovery (Empathy), retention/value (Stickiness), referral growth (Virality), monetization (Revenue), or repeatable expansion (Scale).
2. Write one risky assumption. Example: “New users who finish onboarding will return within seven days.”
3. Pick one OMTM. Choose a metric that can directly strengthen or weaken that assumption. Avoid selecting a metric merely because it is easy to collect.
4. Establish a baseline and threshold. Record the current value and decide what result would be good enough before the experiment begins.
5. Segment the result. Compare at least one cohort, acquisition source, customer type, or time period so an aggregate does not hide important differences.
6. Make one decision. At the end of the review period, choose a concrete action: continue, change the experiment, change the segment, or change the metric.
Reflection and rereading prompts
- Which metric on your current dashboard looks impressive but would not change a decision if it moved 20%?
- What is the riskiest assumption in your product today?
- Are you optimizing acquisition before proving retention?
- Which cohort would reveal whether the product is improving rather than merely accumulating users?
- What threshold would you define before the next experiment begins?
- Is your OMTM a cause, a proxy, or only a lagging result?
- What business-model characteristic makes a generic industry benchmark misleading for you?
- If the current OMTM improves, what is the next likely bottleneck?
- Which decision still requires judgment even after the data is known?
Reading note
Key ideas
- One Metric That Matters (OMTM): A company will track many numbers, but at a given stage one metric should receive disproportionate attention because it best represents the current bottleneck or risk. OMTM creates focus; it is meant to change as the company progresses.
- Know your stage before choosing your metric: The five Lean Analytics stages are Empathy, Stickiness, Virality, Revenue, and Scale. Measuring the wrong stage can lead to premature optimization—for example, buying acquisition before proving retention.
- Empathy comes before growth: Early work is largely qualitative. The team needs to understand whether a real group of people has a meaningful problem and whether the proposed solution addresses it. Interviews and direct observation matter before sophisticated dashboards.
- Stickiness means proving that people get value repeatedly: Retention, engagement, frequency, activation, and cohort behavior become more important than raw signups. If users do not return, accelerating acquisition usually magnifies the leak.
- Virality is a growth mechanism, not a substitute for value: Referral loops and viral coefficients matter only after the product is useful enough to retain people. Growth that brings users into a weak product can create impressive top-line charts without a durable business.
- Revenue must be economically healthy: At the Revenue stage the company has to prove that monetization can work consistently. Metrics such as customer acquisition cost, customer lifetime value, margins, conversion, and time to recover acquisition spend become central.
- Scale changes the optimization problem: After a business has a repeatable model, the questions shift toward efficiency, predictability, market expansion, channels, operations, and sustainable growth. A metric that was useful for discovery may no longer be the right operating metric.
- Good metrics are comparative and actionable: Ratios, rates, segments, and cohorts often reveal more than cumulative totals. A useful metric should help a team decide what to do differently rather than simply report that activity occurred.
- Cohort analysis separates product improvement from accumulated growth: Comparing users who started in different periods can show whether retention or behavior is actually improving. A rising total user count can hide deteriorating cohort quality.
- Set a line in the sand before interpreting results: Define the threshold that would count as good enough before the experiment ends. Otherwise teams can rationalize almost any result after seeing the data.
Deep analysis
The real problem Lean Analytics solves
Most organizations do not suffer from a lack of data. They suffer from a lack of decision focus. A startup can instrument every click and still avoid the fundamental question of whether it is solving a real problem. A product team can celebrate rising monthly users while retention is weakening. A marketing team can reduce acquisition cost while attracting customers whose lifetime value is even lower.
Lean Analytics tries to stop this form of self-deception. The book’s deepest contribution is the connection between business model + stage + metric. A metric has meaning only in relation to the question the company is trying to answer.
1. One Metric That Matters is a focusing device
The OMTM is often misunderstood as “a business needs only one KPI.” That is not the argument. Companies should retain diagnostic data and monitor important health metrics. The OMTM is the number that receives the most attention because it best captures the current constraint.
For an early product, that might be the percentage of interviewed customers who experience a particular problem with sufficient intensity. Later it might be weekly retention, invitation rate, conversion to paid, contribution margin, or another metric tied to the stage. The point is to create organizational alignment: if everyone knows the current question and the metric that answers it, teams are less likely to optimize unrelated numbers.
OMTM also creates a natural review cycle. Once the metric crosses a meaningful threshold, the company asks whether the dominant risk has changed. Focus moves; it is not permanently attached to one dashboard.
2. Stage One — Empathy: prove the problem deserves to exist
In the Empathy stage the team is learning about customers, not proving that its solution is brilliant. Qualitative evidence is especially valuable because the unknowns are still broad. Interviews, observation, problem discovery, and early solution tests help determine whether the proposed problem is real, frequent, painful, and shared by a sufficiently useful market.
This stage is where many “data-driven” teams make a category error. They want statistically clean product metrics before they have a product or stable behavior to measure. Lean Analytics instead treats conversations and structured qualitative evidence as legitimate data. The important discipline is to avoid leading questions and to look for behavior rather than polite enthusiasm.
3. Stage Two — Stickiness: prove that the product creates repeat value
Once a product exists, the key question becomes whether users actually derive enough value to return. This is where retention and engagement matter. Signups alone can be a vanity metric because they measure entry, not sustained value.
Cohort analysis is especially powerful here. Suppose total weekly active users increase for four months. That looks healthy. But if each new cohort retains worse than the previous one and growth is maintained only by spending more on acquisition, the product may be deteriorating. Cohorts make the hidden pattern visible.
Stickiness is also contextual. A payroll product, a daily social app, and an annual tax tool should not have the same expected frequency. The correct metric reflects the natural usage cycle of the product.
4. Stage Three — Virality: understand whether users create more users
Virality can dramatically reduce acquisition friction, but the authors place it after stickiness for a reason. If a product has poor retention, stronger referral loops simply send more people into a leaky system.
Useful viral measures include how many invitations users send, the fraction accepted, the viral coefficient, and the time required for a viral cycle. But the number should connect to a real mechanism. A one-time referral promotion is not the same as a product whose value naturally increases when more participants join.
The broader lesson is that growth metrics should explain why growth happens. A rising chart without a mechanism is fragile because the team does not know which behavior to preserve.
5. Stage Four — Revenue: prove a repeatable business, not just a product
At the Revenue stage the company needs evidence that it can convert product value into an economically sustainable model. Gross revenue is useful, but it can still mislead. If a company spends more to acquire customers than those customers generate, growth can destroy value.
Customer acquisition cost (CAC), customer lifetime value (LTV), gross margin, conversion rate, churn, revenue per customer, and payback time are examples of metrics that help explain economics. Which one becomes the OMTM depends on the model and the current bottleneck.
A particularly useful insight is that “making money” and “having a scalable business” are different. A business may have customers but still depend on founder effort, unusually cheap acquisition, unsustainable discounts, or long payback periods. Revenue-stage analytics should expose those dependencies.
6. Stage Five — Scale: optimize a model that has already earned the right to grow
Scale is not merely “more traffic.” It means the company has enough evidence about product value and economics to widen the system. The challenge becomes finding additional channels, markets, segments, operational capacity, and efficiency without breaking what worked at smaller size.
At this point a single startup-wide metric may become less sufficient because the organization has multiple mature functions. Still, the OMTM principle remains useful inside teams: identify the current constraint rather than allowing every department to optimize a local number that may conflict with the whole business.
7. Business model changes the meaning of “good”
Lean Analytics compares several business-model patterns because the same metric does not carry the same meaning everywhere. A SaaS product cares deeply about recurring revenue, churn, expansion, and acquisition economics. An e-commerce business must understand conversion, repeat purchase, average order value, and margins. A marketplace has to balance supply and demand, liquidity, transaction frequency, and take rate. Media and user-generated-content businesses have different engagement and monetization structures.
This is why generic benchmark lists can be dangerous. A “good” conversion rate is not meaningful without knowing the model, price point, acquisition channel, customer intent, and stage.
8. Vanity metrics versus actionable metrics
Vanity metrics are not necessarily false. They are often incomplete. Total registered users, cumulative downloads, pageviews, or followers can legitimately increase while the underlying business weakens. Their main problem is that they are easy to celebrate and hard to act on.
An actionable metric is connected to a decision. If weekly retention falls from one cohort to the next, a product team can investigate onboarding or product value. If trial-to-paid conversion differs sharply by acquisition channel, marketing and product teams can change channel mix or positioning. If CAC payback becomes too long, the business can adjust acquisition, price, margin, or retention.
9. Leading and lagging indicators
Lagging indicators tell you what has already happened—revenue, churn, or closed sales, for example. Leading indicators may provide earlier evidence about what is likely to happen next, such as activation behavior that historically predicts retention.
The trap is to treat correlation as certainty. A leading indicator is useful only while its relationship with the desired outcome remains reliable. Teams should periodically test whether the proxy still predicts what they care about.
10. Why cohorts beat aggregates
Aggregates mix users with different histories. A product may appear to have 40% retention because long-time loyal users are combined with new users who are leaving rapidly. Cohorts group people by a shared start time or characteristic, making change easier to see.
This is one of the book’s most transferable practices. Cohort thinking applies beyond startups—to email engagement, education, subscriptions, marketplaces, communities, and even content platforms. Whenever a total number combines populations exposed to different versions or conditions, cohort analysis can reveal whether the system is actually improving.
11. Draw a line in the sand
Before running an experiment, decide what result would be strong enough to justify the next step. This protects the team from changing the interpretation after seeing the outcome.
The threshold does not have to be perfect. It can come from industry benchmarks, comparable products, historical baselines, unit economics, or explicit management judgment. Its value is epistemic discipline: the team knows in advance what it is trying to learn.
If the result misses the line, the right response is not automatically to abandon the product. It is to investigate why: wrong segment, weak value proposition, poor execution, inappropriate threshold, or false assumption. The result creates a better question.
12. Data-driven versus data-informed
A mature reading of Lean Analytics is not “numbers should make every decision.” Data can describe behavior but may not capture strategy, ethics, brand, long-term bets, or phenomena for which the sample is too small. The practical goal is to become data-informed: use measurement to challenge intuition without pretending that measurement eliminates judgment.
Founders still choose which market to pursue, what kind of company to build, and which trade-offs are acceptable. Analytics improves those decisions by making assumptions visible.
Where the framework is strongest
Lean Analytics is especially useful when a team is overwhelmed by dashboards or is arguing about priorities. The stage framework gives the conversation order. OMTM makes the bottleneck explicit. Cohorts and comparative metrics reduce the seduction of cumulative growth charts. Lines in the sand reduce post-hoc rationalization.
Where to be cautious
The five stages are a model, not a law of nature. Real companies may revisit earlier stages, run multiple products, monetize before virality, or operate in markets where regulation, enterprise sales cycles, seasonality, or network effects change the sequence. A mature organization can have different products in different stages at the same time.
Benchmarks can also become false precision. What matters is whether the chosen metric links to the economics and behavior of the specific product. A team should use the framework to improve questions, not to force reality into a template.
A practical synthesis
The book can be compressed into a recurring loop:
- Identify the riskiest business assumption.
- Determine the stage and business model that shape the question.
- Select the metric that best represents progress on that risk.
- Establish a baseline and a target before acting.
- Run a focused experiment.
- Compare cohorts or segments where aggregate data could mislead.
- Decide whether to keep, change, or reject the hypothesis.
- When the bottleneck changes, change the OMTM.
That loop is more valuable than memorizing a long list of startup metrics. It turns analytics from reporting into a system for learning.
Practical application
Apply Lean Analytics to a real project
1. Name the stage. Decide whether your biggest uncertainty is problem discovery (Empathy), retention/value (Stickiness), referral growth (Virality), monetization (Revenue), or repeatable expansion (Scale).
2. Write one risky assumption. Example: “New users who finish onboarding will return within seven days.”
3. Pick one OMTM. Choose a metric that can directly strengthen or weaken that assumption. Avoid selecting a metric merely because it is easy to collect.
4. Establish a baseline and threshold. Record the current value and decide what result would be good enough before the experiment begins.
5. Segment the result. Compare at least one cohort, acquisition source, customer type, or time period so an aggregate does not hide important differences.
6. Make one decision. At the end of the review period, choose a concrete action: continue, change the experiment, change the segment, or change the metric.
Continue exploring
Lean Analytics FAQs
Who wrote Lean Analytics?
Lean Analytics was written by Alistair Croll and Benjamin Yoskovitz.
What is Lean Analytics about?
Lean Analytics by Alistair Croll and Benjamin Yoskovitz is a practical framework for using data to reduce the biggest uncertainty in a startup or new product. Its central argument is not “measure everything.” It is the opposite: understand your business model, identify the stage you are actually in, and focus the organization on the One Metric That Matters (OMTM) for the question that matters now. The book combines Lean Start…
What are the key ideas in Lean Analytics?
One Metric That Matters (OMTM): A company will track many numbers, but at a given stage one metric should receive disproportionate attention because it best represents the current bottleneck or risk. OMTM creates focus; it is meant to change as the company progresses. Know your stage before choosing your metric: The five Lean Analytics stages are Empathy, Stickiness, Virality, Revenue, and Scale.
Is this the full text of Lean Analytics?
No. This Booknomics page is an independent study summary and analysis of Lean Analytics; it is not the full text of the original book.
Related book summaries
- Blitzscaling — Reid Hoffman and Chris Yeh
- Choose Your Enemies Wisely — Patrick Bet-David, Greg Dinkin
- Crushing It! — Gary Vaynerchuk
- Disciplined Entrepreneurship — Bill Aulet