Lean Analytics
Which metric actually governs each business-model type.
Summary
Croll and Yoskovitz give a framework for choosing and using metrics in a startup, on the argument that most companies measure the wrong things and that the correct metric depends on the business model and on the stage.
Their organising concept is the One Metric That Matters, meaning that at any given stage a company should focus on a single number, because attention is limited and multiple targets produce no decisions. They pair this with the stage framework: empathy, in which the company establishes that a real problem exists; stickiness, in which it establishes that the product retains users; virality, in which it establishes that users bring others; revenue, in which it establishes the economics; and scale, in which it grows.
The business model chapters are the practical core. They give six models, which are ecommerce, software as a service, free mobile applications, media, user-generated content and two-sided marketplaces, and for each they specify the metrics that matter, the benchmark values a company should compare against, and the specific ways each model fails.
They also cover the distinction between vanity metrics and actionable ones, cohort analysis as the correct method for measuring retention, the requirement that a metric be comparative and understandable, and the practice of drawing a line in the sand, meaning stating in advance what number would count as success before running the experiment. The benchmark values are from around 2013 and have shifted substantially. The book is long for its content and repeats the framework. The model-specific metric lists and the benchmark discipline are what practitioners use.
Key takeaways 8
- A company should focus on a single metric at any given stage, because multiple targets produce no decisions.
- The correct metric depends on the business model and on the stage, so borrowing metrics from another company misdirects effort.
- Startups pass through stages of establishing a real problem, retention, referral, revenue and scale, and metrics differ at each.
- Cohort analysis rather than aggregate totals is the correct method for measuring retention, since aggregates conceal deterioration.
- A metric must be comparative, understandable and capable of changing behaviour, or it is not worth collecting.
- Vanity metrics including cumulative totals rise regardless of performance and should be excluded from any decision.
- A target should be stated before running an experiment, so that the result cannot be reinterpreted after the fact.
- Retention should be established before growth is pursued, since acquiring users into a product that does not retain them wastes the spending.
Who should read it
Founders and product teams choosing what to measure, particularly those who need model-specific metrics rather than general advice.
Who recommended it 3
Similar books 6
Chosen by shared topics, the entries that hold them, and the people who recommended both.
When and how to put speed before efficiency to capture a market before rivals.
Recommended by: Anthony Pompliano, Bill Gates, Biz Stone and 7 others
Operating manual for scaling a company from fifty to thousands
Recommended by: Balaji S. Srinivasan, Matt Mullenweg and 10 others
Owner-operator economics: labour efficiency and the four keys.
Recommended by: Nat Eliason
Build, measure, learn: test what customers want with small experiments.
Recommended by: Nat Eliason, Sheryl Sandberg, Andrew Chen and 17 others
The foundational text of the modern PM role. Cagan's central argument — that most companies build the wrong things because they treat product as a delivery function rather than a discovery function — reframes everything that follows. The chapter on product discovery versus product delivery is worth the price of the book.
Recommended by: Anna Cheng, Marty Cagan
Recommended by: Mohnish Pabrai
Where it sits in the reference 4
Topics