Eric Ries’s The Lean Startup: How Today’s Entrepreneurs Use Continuous Innovation to Create Radically Successful Businesses is the management book that popularised the vocabulary product development now uses: the minimum viable product, the pivot, validated learning and the build-measure-learn loop. Published in 2011 by a software entrepreneur who had watched well-funded companies build things nobody wanted, it argues that a new venture, inside a start-up or a large organisation, is an experiment whose purpose is to learn what customers will actually do, that most of the effort in traditional product development produces no learning, and that the discipline of running small, fast, measured experiments is the only reliable way to find out. For an internal auditor the book is more relevant than its title suggests, because its central chapter is about accounting, specifically about how to tell whether an innovation programme is making progress when conventional measures cannot, and because it insists, in so many words, that the metrics must be auditable. This guide summarises the book without reproducing it, sets out its method, maps it to the audit of innovation and change, and ends with a worked example of a food group’s audit function assessing its innovation programme with the book’s innovation accounting.
In this guide
- The book in one page
- Who Eric Ries is, and how the book came about
- The structure: vision, steer, accelerate
- The method: build, measure, learn
- Innovation accounting: the chapter for auditors
- Actionable, accessible, auditable: the three requirements for a metric
- The pivot catalogue, the engines of growth and the five whys
- The nine ideas worth keeping
- The vocabulary: fourteen terms defined for practitioners
- What it means for internal audit: auditing innovation and change
- The criticisms, and what the book leaves out
- Editions, reading time and how to read it
- A one-hour discussion guide for an audit team
- Worked example: Pennine Foods audits its innovation programme
- Frequently asked questions
- Related guides
The book in one page
| Question | The answer, in our words |
|---|---|
| What is a start-up, in the book’s sense? | Any human institution creating a new product or service under conditions of extreme uncertainty, whether a two-person company or a team inside a multinational; the defining feature is that nobody yet knows whether customers want the thing. |
| What is the problem? | That conventional management, with its plans, forecasts and milestones, assumes the uncertainty away; teams execute a plan to build a product, measure progress by features delivered and budget spent, and learn that nobody wanted it only at launch. |
| What is the method? | Treat every assumption about customers as a hypothesis; build the smallest thing that can test it (the minimum viable product); measure what customers actually do; learn and decide whether to persevere with the strategy or pivot to a new one; repeat as fast as possible. |
| What is innovation accounting? | A way of measuring progress when revenue and profit cannot: establish a baseline of how customers behave, tune the product toward the ideal, and decide at defined points whether the trend justifies continuing. Its metrics must be actionable, accessible and auditable. |
| Where does it come from? | Lean manufacturing (small batches, the five whys, stopping the line), customer development, and the author’s own failures and successes at the software company IMVU. |
| Why read it as an auditor? | Because innovation spending is large and rarely audited well, because the book provides a control framework for it that the audit profession has mostly ignored, and because its demand for auditable metrics and its warnings about vanity metrics are an audit standard stated by a founder. |
| Reading time | About eight hours for the 320-page paperback; the innovation accounting chapters are the ones an auditor should read slowly. |
Who Eric Ries is, and how the book came about
Ries co-founded IMVU, a virtual-world company, in 2004, after two earlier start-ups had failed, and at IMVU he and his colleagues adopted the practices of releasing early, measuring everything and changing direction on evidence that the book describes. He began writing about the method on a blog in 2008, borrowed the word lean from the Toyota production system and the customer-development ideas of his mentor Steve Blank, and published the book with Crown Business in September 2011. It became a bestseller, a management movement with conferences and consultancies, and a vocabulary that large companies adopted, often, as Ries later complained, without the discipline. His 2017 book The Startup Way applied the method to established organisations, and his later work on the Long-Term Stock Exchange addressed governance. The book is a practitioner’s manual with case studies, written for founders and managers, and its relevance to audit is an accident of its honesty about measurement.
The structure: vision, steer, accelerate
| Part | What it covers | What it contributes |
|---|---|---|
| Part one: vision | The definition of a start-up, the idea of validated learning, the experiment as the unit of progress, and why effort that produces no learning is waste | The frame: a new venture is a learning machine |
| Part two: steer | The build-measure-learn loop, the leap-of-faith assumptions, the minimum viable product, innovation accounting and the pivot | The method, including the chapters an auditor needs |
| Part three: accelerate | Small batches, engines of growth, adaptive organisations and the five whys, and how the method works inside large companies | The operating practices |
The method: build, measure, learn
The loop is the book’s organising device. A team starts with a vision and a set of assumptions about customers, of which two are leaps of faith: the value hypothesis, that the product delivers value to customers, and the growth hypothesis, that customers will come. Rather than building the product the vision describes, the team builds the minimum thing that can test the riskiest assumption, which may be a landing page, a video, a concierge service performed by hand, or a stripped-down version of the product. It then measures what customers do, in a way that isolates the effect of the change, and learns whether the assumption held. The learning drives the next build. The loop’s speed is the metric that matters: a team that completes it weekly learns more in a quarter than one that completes it annually, whatever the quality of the two visions. The change management audit guide and the Phoenix Project summary describe the same logic applied to software delivery: small batches, fast feedback, and the recognition that the size of a release is itself a risk.
| Stage | What happens | The failure it is designed to prevent | The question an auditor asks |
|---|---|---|---|
| Build | The minimum viable product that tests the riskiest assumption | Building the whole product before knowing anyone wants it | What assumption does this build test, and is it the riskiest? |
| Measure | Metrics that show how customers behave, by cohort, with a baseline | Vanity metrics that rise regardless of what the team does | Are the metrics actionable, accessible and auditable? |
| Learn | Validated learning: a conclusion about customer behaviour supported by data | Learning that is asserted rather than demonstrated | What did the team conclude, and what evidence supports it? |
| Pivot or persevere | A decision, at a defined point, to keep the strategy or change it | Perseverance by default; the project that cannot be stopped | When was the last decision point, and what would have triggered a pivot? |
Innovation accounting: the chapter for auditors
The chapter on innovation accounting addresses the problem that makes innovation hard to govern: conventional accounting measures a new venture at zero, revenue nil, costs rising, for as long as it takes to find out whether it will work, and so provides no signal about whether the money is being spent well. Ries proposes three learning milestones. First, establish the baseline: run the minimum viable product and measure how customers actually behave, conversion, retention, usage, whatever the business model depends on, however poor the numbers. Second, tune the engine: every subsequent change is an experiment whose purpose is to move those numbers toward the ideal the business model requires, and progress is the movement. Third, pivot or persevere: at defined intervals the team and its sponsors examine whether the tuning is moving the numbers fast enough to reach the model, and if not, change the strategy. The milestones give a sponsor, a board or an auditor a way to evaluate a venture that has no profit: is the baseline measured, is it improving, and is the improvement on a trajectory that could succeed? The strategic risk guide treats innovation portfolios as a strategic exposure; innovation accounting is the control framework that exposure lacks.
Actionable, accessible, auditable: the three requirements for a metric
The book’s standard for a metric is the passage an auditor should keep. A metric is actionable if it demonstrates clear cause and effect: it must be possible to say what the team did to move it, which rules out gross totals that rise with time and marketing spend regardless of product changes, the vanity metrics the book attacks, and favours cohort analysis and split tests that isolate the effect of a change. It is accessible if the people who need to act on it can understand it without a data specialist: reports expressed in people and actions rather than abstractions, available to the whole team. And it is auditable if its figures can be traced to the raw data and reproduced: Ries notes that people whose numbers disappoint tend to blame the data, and his rule is that a report must be traceable to the individual customers behind it, so that a sceptic can check it, if necessary by talking to those customers. The third requirement is unusual in management writing and exact in its demand: a management reporting control, stated by a founder as a condition of being believed.
| Requirement | What it means | The vanity-metric failure | The audit test |
|---|---|---|---|
| Actionable | Clear cause and effect between the team’s actions and the number | Total registered users rising while the product does nothing | Can the team name the change that moved the metric? Is the analysis by cohort? |
| Accessible | Understandable by the people who act on it | A dashboard only the analyst can read | Interview a team member on what the report says and what they would do about it |
| Auditable | Traceable to raw data and reproducible | A figure nobody can reconstruct | Reperform one figure from the underlying records; ask for the data behind a disputed one |
The pivot catalogue, the engines of growth and the five whys
Three further parts of the method matter for an auditor because each is a control in disguise. The first is the pivot catalogue. Ries lists the forms a change of strategy can take, among them the zoom-in pivot (a single feature becomes the product), the zoom-out (the product becomes a feature of something larger), the customer segment pivot (same product, different buyer), the customer need pivot (same buyer, different problem), the platform pivot, the business architecture pivot (high margin and low volume to the reverse), the value capture pivot (a different way of charging), the engine of growth pivot, the channel pivot and the technology pivot. The catalogue is a vocabulary for a decision that most programmes never record, and its use in a steering committee is evidence that the decision was considered. The second is the engine of growth, the mechanism by which a product acquires customers sustainably, which Ries reduces to three. Each engine has its own metric, which is why a dashboard built on totals cannot show whether the engine works.
| Engine | How customers arrive | The metric that matters | The vanity metric that hides it |
|---|---|---|---|
| Sticky | Existing customers stay and keep paying; growth is the margin by which acquisition exceeds churn | Retention and churn rate by cohort | Total customers |
| Viral | Existing customers bring new ones as a side effect of using the product | The viral coefficient: new customers per existing customer | Sign-ups |
| Paid | Advertising and sales bought with the margin earned on each customer | Lifetime value against cost of acquisition | Revenue growth while the acquisition cost is rising faster |
The third is the five whys, which Ries adapts from Toyota as the mechanism by which an organisation learns from its failures without blaming its people. When something goes wrong the team asks why, and why again, until a process cause is reached, and then invests in a fix in proportion to the size of the problem: a small problem earns a small fix, a repeated one a larger. His rule that a small problem earns only a small fix, with more invested if it recurs, is a sensible discipline for an audit function tempted to recommend a new control for every finding, and it echoes the proportionality that the root cause guide builds into its method. The three together, the recorded pivot, the engine-specific metric and the proportional root-cause fix, are the operating practices that make the loop more than a slogan; a programme that has adopted the vocabulary without them has adopted nothing.
The nine ideas worth keeping
- A new venture is an experiment, and its progress is learning. Effort that produces no validated learning is waste however much it delivers.
- The riskiest assumption first. Test what, if wrong, kills the venture, before building what, if right, decorates it.
- Minimum viable means minimum. The smallest thing that tests the assumption; the book’s examples are a video, a landing page and a founder doing by hand what the product will later automate.
- Vanity metrics are a control failure. Numbers that rise regardless of what the team does report success without evidence of it; cohorts and controlled comparisons are the alternative.
- Metrics must be auditable. Traceable to raw data, reproducible, available to the sceptic; a founder’s statement of a management reporting control.
- Innovation accounting replaces profit where profit cannot yet exist. Baseline, tune, pivot or persevere: a sponsor can govern a venture with no revenue if those three are in place.
- The pivot is a decision point, not a failure. Scheduled, evidenced and recorded; the venture that never pivots has usually never been measured.
- Small batches reduce risk. From manufacturing through software to organisational change, the batch size is the risk; the change management audit guide is built on it.
- The five whys find the system, not the person. Ries’s adaptation of Toyota’s root cause method, with the rule of proportional investment in the fix; the root cause guide uses the same discipline.
The vocabulary: fourteen terms defined for practitioners
| Term | What Ries means | What it looks like in an audit |
|---|---|---|
| Validated learning | A conclusion about customers demonstrated by experiment | The evidence behind a project’s claimed progress |
| Build-measure-learn | The experimental loop | The project’s cycle time and what each cycle tested |
| Leap-of-faith assumption | The assumption on which the venture depends | The riskiest hypothesis in the business case |
| Minimum viable product | The smallest build that tests an assumption | Scope discipline; the test of whether the first release was a test |
| Pivot | A structured change of strategy on evidence | The decision point and its record |
| Persevere | Continuing the strategy on evidence | The same decision, the other way |
| Innovation accounting | Baseline, tune, pivot or persevere | The governance framework for ventures without profit |
| Vanity metric | A number that rises regardless of the team’s actions | A reporting control failure |
| Actionable metric | A number with clear cause and effect | Cohort analysis; controlled comparison |
| Cohort analysis | Tracking groups of customers who joined at the same time | The method that makes a metric actionable |
| Split test | Offering different versions to compare behaviour | A controlled experiment; evidence of causation |
| Engine of growth | The mechanism by which customers bring customers: sticky, viral or paid | The growth hypothesis to be tested |
| Five whys | Asking why repeatedly to reach a root cause, with proportional investment in the fix | Root cause analysis |
| Andon cord | Toyota’s mechanism for any worker to stop the line | Authority to stop; see the Checklist Manifesto summary |
What it means for internal audit: auditing innovation and change
Innovation programmes, digital transformations and new-product portfolios absorb large budgets and are audited, when they are audited at all, for project governance: were the milestones met, the budget kept, the steering committee held. The book supplies a different and better question, whether the spending produced validated learning on a trajectory that could succeed, and a set of controls for answering it. The table maps the method to audit practice. The strategic risk guide, the change management audit guide and the analytics evidence guide carry the surrounding procedures.
| Audit activity | What the book implies | Procedure it changes |
|---|---|---|
| Auditing an innovation programme or venture | Progress is learning; the three milestones are the governance | Assess whether a baseline was measured, whether it moved, and whether decision points exist; the programme that reports features delivered and nothing about customers has no governance |
| Reviewing business cases | The leap-of-faith assumptions should be named and tested first | Identify the assumption that, if wrong, kills the case; ask what test has been run on it |
| Testing management reporting | Actionable, accessible, auditable | The three-requirement test on the programme’s dashboard; reperformance of one figure from raw data |
| Evaluating project governance | Pivot-or-persevere decisions, scheduled and evidenced | Read the steering committee minutes for a decision point at which the strategy could have changed; if none, report that |
| Change and release management | Small batches and fast feedback | The change management audit guide‘s pipeline model |
| Root cause analysis | The five whys with proportional investment | The root cause guide |
| Innovation portfolio risk | Ventures that cannot be stopped are the risk | Count the ventures in the portfolio that have passed a decision point in the last year |
| The function’s own methodology | The function is itself a product with customers | Treat a new audit approach as an experiment with a baseline and a measure |
The criticisms, and what the book leaves out
The book has been criticised from inside its own movement and from outside it. Practitioners note that the minimum viable product has been used to justify releasing poor products, that the method suits software with cheap experiments far better than hardware, pharmaceuticals or regulated services where experiments are expensive and sometimes impermissible, and that large companies adopted the vocabulary as theatre, running “lean” programmes with no baseline and no decision points; Ries takes up the last problem himself in The Startup Way. Academics argue that the method’s success stories are survivors and that controlled evidence is thin, which is an objection the book’s own standard of evidence invites; the best-known test, a 2020 randomised trial with 116 Italian start-ups, found that founders taught to frame and test hypotheses performed better and pivoted more. It says little about the governance of experiments on customers, including privacy and consent, which has mattered more since 2011, or about the regulatory constraints that apply to experimenting in financial services and healthcare. For an auditor the limits are a reason to apply the framework selectively, to ventures where experiments are feasible, and to keep the part that applies everywhere: the demand that progress be measured, that metrics be auditable and that a venture face a decision point at which it can be stopped.
Editions, reading time and how to read it
| What | Details | Best for |
|---|---|---|
| Hardback | Crown Business, September 2011 | The first edition |
| Paperback | Currency, a Crown imprint, 2017 | Readers who prefer a paperback |
| Audiobook | Unabridged, read by the author, about nine hours | Works well; the cases are told as stories |
| Reading plan | Part one quickly; the chapters on innovation accounting and on the pivot slowly; the five whys chapter | An auditor with one long evening |
| Companion reading | The Startup Way (2017) for large organisations; the Phoenix Project summary for the delivery side; the Superforecasting summary for the discipline of measured prediction | Placing the method |
A one-hour discussion guide for an audit team
- Ten minutes: the portfolio. List the organisation’s innovation, transformation and new-product initiatives above a budget threshold, and for each name the assumption that, if wrong, would end it.
- Fifteen minutes: the dashboard. Take the largest initiative’s most recent progress report and test each metric for the three requirements. Mark the vanity metrics.
- Ten minutes: the decision point. Ask when the initiative last faced a scheduled decision at which it could have been stopped or redirected, and what the evidence was.
- Fifteen minutes: the reperformance. Choose one headline figure from the report and discuss how the function would reproduce it from raw data, and whether it could.
- Ten minutes: the plan. Decide whether an innovation-accounting review of the portfolio belongs in the next plan and who sponsors it; the strategic risk guide gives the frame.
Worked example: Pennine Foods audits its innovation programme
Pennine Foods plc is a fictional UK-listed food group with £900 million of revenue, a December year-end and a four-person internal audit function. Its innovation programme, £14 million over three years, ran seven new-product and direct-to-consumer initiatives through a steering committee that received monthly reports on milestones and spend. The audit committee asked for a review, and the function, rather than auditing project governance, applied the book’s innovation accounting to each initiative.
| Test | What the function found | Result |
|---|---|---|
| Is the riskiest assumption named and tested? | Five of seven business cases listed market size and margin assumptions; none identified the assumption that would end the initiative if wrong; two initiatives had spent over £2 million before any customer had been asked to buy | Finding: every initiative to name its leap-of-faith assumption and the test run on it; two initiatives paused pending a test |
| Is there a baseline? | The direct-to-consumer initiative reported registered users (118,000) and app downloads; it had never reported repeat purchase by cohort | Finding: cohort reporting introduced; the first cohort analysis showed 9 percent of users buying twice, against a business case requiring 35 percent |
| Are the metrics actionable, accessible, auditable? | Of fourteen metrics on the committee dashboard, nine were totals that rose with spend; the function could not reproduce the “engagement score” from raw data and the analytics team could not explain its formula | Finding: the score removed; the function reperformed the remaining figures, two of which were wrong by more than 20 percent |
| Is there a decision point? | The steering committee had never stopped or redirected an initiative; the minutes recorded no occasion on which the question was put | Finding: quarterly pivot-or-persevere decisions with stated criteria; at the first one, two initiatives were stopped and one pivoted from a product to a licensing model |
| Are batches small? | The flagship product initiative planned a single national launch after eighteen months of development; a regional pilot had been proposed and rejected as “sending the wrong signal” | Observation: a regional pilot adopted; it found a packaging problem that the national launch would have found in every store |
| Are root causes found? | A failed launch the previous year had been attributed to the retailer’s buyer; the five whys, run by the function with the team, reached a forecasting process that had never been validated | Observation: the root cause guide‘s method adopted for launch reviews |
The review produced four findings and two observations, and its effect on the programme was larger than its findings: two initiatives stopped, one pivoted, £3.1 million of planned spend released, and a steering committee that for the first time received numbers it could check and faced decisions it could make. The commercial director’s remark in the closing meeting, that the function had audited the programme as if it were a business rather than a project, is the book’s argument in one sentence. Innovation accounting is a control framework; the book called it something else because its author was writing for founders, and the audit profession has been slow to notice.
Frequently asked questions
Is The Lean Startup only for technology start-ups?
No; the book’s own cases include Intuit, a large software company, as well as start-ups, and the method applies wherever a new product, service or process is launched under uncertainty. It applies best where experiments are cheap and worst where they are expensive or restricted, and an auditor should apply it selectively.
What is innovation accounting?
A way of governing a venture that has no profit yet: measure a baseline of customer behaviour, tune the product toward the business model’s requirements, and decide at defined points whether the trajectory justifies continuing. It replaces milestones and spend as the measure of progress.
Why does the book say metrics must be auditable?
Because teams lose arguments and make bad decisions on figures nobody can check; Ries requires that any report be traceable to raw data and reproducible, available to the sceptic. It is a management reporting control stated by a founder, and the analytics evidence guide describes the audit version.
What is a vanity metric?
A number that rises regardless of what the team does, total users, cumulative downloads, page views, and so reports success without evidence of it. The alternative is cohort analysis and controlled comparison, which show cause and effect.
How does the book relate to The Phoenix Project?
Both come from the lean tradition and both argue that small batches and fast feedback reduce risk; this book is about what to build and whether it works, the other about how to deliver it. The Phoenix Project summary covers the delivery side.
Where does the book sit in the reading list?
In the technology and change group of the internal audit reading list, with The Phoenix Project and the Checklist Manifesto.
Related guides
- Strategic Risk: A Comprehensive Guide — innovation portfolios as a strategic exposure
- How to Audit IT Change Management — small batches and fast feedback in delivery
- Documenting Analytics as Audit Evidence — the auditable metric, from the audit side
- Root Cause Analysis for Audit Findings — the five whys, applied
- The Phoenix Project: Summary and What It Means for IT Auditors — the delivery side of the same tradition
- Superforecasting: Summary and Lessons for Auditors — measured prediction as a discipline
- The Best Books for Internal Auditors — where this book sits among the others
- All book summaries — the full set of summaries for auditors
Leave a Reply