Innovation & product
Lean Canvas
A one-page adaptation of the Business Model Canvas for startups, replacing four infrastructure boxes with Problem, Solution, Key metrics and Unfair advantage, so founders can document, rank and test the riskiest assumptions in a business idea instead of writing a plan.
Also known as Lean startup canvas, Maurya canvas. First set out by Ash Maurya in 2010; the primary source is cited in full below.
Where this is contested
The Lean Canvas is an open adaptation of Osterwalder and Pigneur's Business Model Canvas, made under its Creative Commons BY-SA licence; Maurya documents the derivation and his reasoning for the four substituted boxes, but which canvas better serves early-stage work remains an argued point between the two schools.
- Format
- Canvas
- Level
- Product · Business unit
- Best for
- Evaluate options · Understand customers · Assess risk
- Decision stage
- Explore options · Plan
- Difficulty
- Introductory
- Time to apply
- Twenty minutes for a first sketch; the useful work is the weeks of testing that follow, with the canvas revised as evidence arrives.
Plate · The model
The components
Problem
The top one to three problems your early adopters have, stated in their language, plus the existing alternatives they use today. Replaces Key partners on Osterwalder's grid.
Signals of strength
Problems a named customer has actually described · Existing alternatives listed honestly, including 'spreadsheet' and 'do nothing' · Fewer than four problems
Solution
The minimum feature set that addresses the top problems. Deliberately a small box: Maurya's point is that founders over-invest here first, and the solution should be bound to tested problems. Replaces Key activities.
Signals of strength
Each feature maps to a listed problem · Small enough to build a demo or MVP quickly · Written after the Problem box, never before
Key metrics
The handful of numbers that tell you how the business is doing, typically stages of the customer lifecycle: acquisition, activation, retention, revenue, referral. Replaces Key resources.
Signals of strength
Actionable metrics tied to customer behaviour · One metric identified as the current focus · No vanity counts such as cumulative sign-ups
Unique value proposition
The single, clear, compelling message that states why you are different and worth attention, aimed at the early adopter. Sits at the centre of the canvas, as Value propositions does on the BMC.
Signals of strength
Understandable by a stranger in seconds · Speaks to the finished-story benefit · Distinct from what the existing alternatives promise
Unfair advantage
Something that cannot easily be copied or bought: insider information, a dream team, personal authority, network effects, community. Maurya's rule is that a real one is rare at the start, and an honest blank beats a platitude. Replaces Customer relationships.
Signals of strength
Would survive a well-funded copycat reading your canvas · Not 'first mover', 'passion' or 'features' · Often blank on day one, filled in as one is earned
Channels
Your free and paid paths to customers: content, search, outbound, partnerships, app stores. Early on the job is finding any repeatable path to early adopters, and scalability comes later.
Signals of strength
At least one channel you can start using this week · Cost per acquisition guessed, then measured · Matched to where early adopters already are
Customer segments
The target customers and users, narrowed to the early adopters most likely to buy first. On a two-sided model, each side is named, and Maurya advises a separate canvas per segment where they differ materially.
Signals of strength
Early adopter described specifically enough to find one · Buyer and user distinguished where they differ · One segment per canvas
Cost structure
Fixed and variable costs of operating the model: people, hosting, acquisition. Paired with Revenue streams to give a first read on burn and break-even.
Signals of strength
Covers the cost of running the tests, not just the eventual business · Customer acquisition cost included · Rough break-even point calculated
Revenue streams
How the model makes money: price, billing model, lifetime value. Maurya argues for charging from day one where possible, because price is part of the product and payment is the strongest validation.
Signals of strength
A price, even provisional, rather than 'free for now' · Revenue logic consistent with the segment's willingness to pay · Lifetime value compared against acquisition cost
When it earns its keep
- You have a startup or new-product idea and need to get the whole business model out of your head and onto one page in under an hour.
- You want a shared, revisable artefact for co-founders or sponsors that makes the assumptions explicit enough to argue about and test.
- You are choosing between several ideas or customer segments and want to compare them on equal terms, one canvas per segment.
- You are entering a discovery phase and need to decide which assumption, if wrong, kills the business, so you can test that one first.
And when it doesn't
- You are mapping an established business with real partners, channels and operations. Osterwalder's Business Model Canvas keeps the infrastructure boxes for good reason.
- The business depends fundamentally on partnerships, regulation or key resources, for instance in energy, pharma or infrastructure. The boxes Maurya removed are exactly the ones you need.
- You need an operating or financial plan. The canvas holds hypotheses at one-page resolution; investors and banks will still want the numbers behind it.
- The canvas is being filled in to decorate a pitch deck after the decisions are made. It is a thinking and testing tool, and it dies laminated.
How to run it
Before starting, gather the inputs the analysis depends on:
- A specific early-adopter customer segment in mind, even if it is a guess to be corrected.
- A first articulation of the top one to three problems that segment has, in the customer's language.
- Rough numbers for costs and pricing, honest enough to expose whether the model could ever work.
- A willingness to leave boxes blank rather than fill them with plausible padding; blank is information.
- 1
Sketch the whole canvas in one sitting
Maurya's guidance is twenty minutes for the first pass. Speed is the point: a fast sketch captures what you actually believe, while a laboured one drifts into what sounds impressive. Leave blank any box you cannot answer.
- 2
Start with Customer segments and Problem
Name the specific early adopter, then the top three problems as that customer would state them. Everything else on the canvas is downstream of these two boxes, which is why Maurya put Problem where Osterwalder had Key partners.
- 3
Write the Unique value proposition
A single clear message stating why you are different and worth attention, aimed at the early adopter rather than the mainstream. A useful test is whether it names the finished-story benefit, not the feature.
- 4
Fill the remaining boxes without polishing
Solution gets the minimum feature set for the top problems. Channels, Revenue streams and Cost structure get first honest guesses. Key metrics gets the handful of numbers that would tell you the model is working. Unfair advantage can stay blank; most real ones are earned later.
- 5
Rank the risks
The canvas is a hypothesis log, so ask which box, if wrong, kills the business soonest. For most startups that is Problem or Customer segments, which is why building the Solution first is usually the wrong opening move.
- 6
Test, then update the canvas
Run problem interviews or small experiments against the riskiest box, and revise the canvas with what you learn. Date each version. A canvas that has never changed is a record of not learning.
Reading the result
A dated, one-page statement of the business model as a set of hypotheses, with the riskiest assumptions identified and a testing order agreed. Its value compounds across versions: the sequence of canvases is a record of what the team learned.
- Read Problem and Customer segments first; if those two boxes are weak, the rest of the canvas is decoration.
- Look at what is blank. An empty Unfair advantage box is honest; an empty Key metrics box means the team cannot yet say what success looks like.
- Compare versions over time. The boxes that keep changing are where the learning is happening; boxes that never change are either solid or unexamined.
A worked example
A dog-walking marketplace sketches and stress-tests its plan A
Two founders in Guildford are building WagRoute, a marketplace app connecting vetted local dog walkers with owners who commute into London. Before writing any code they draft a Lean Canvas for the owner side, timeboxed to twenty minutes, then use it to decide what to test first.
- Problem
- Commuting owners cannot get reliable midday walks; existing alternatives are a patchwork of neighbours, a teenager after school, or an established franchise with a waiting list. Trust is the stated anxiety: strangers holding house keys.
- Customer segments
- Early adopter: dual-income households within 15 minutes of a mainline station, one dog, no garden access at midday. Walkers are the other side of the marketplace and get their own canvas.
- Unique value proposition
- First draft: 'A walker your dog already knows, every working day.' The differentiator is the same vetted walker each time with GPS-tracked walks, against the franchise model's rotating staff.
- Solution
- Minimum set: walker vetting and repeat-matching, booking for a standing weekly slot, walk report with route map. Key-exchange logistics deliberately excluded from version one.
- Channels
- Vets, groomers and station-car-park advertising for owners; existing self-employed walkers recruited directly. No paid social until a channel proves repeatable.
- Revenue streams
- Commission of 20 per cent on a 14 pound walk. Owners pay in-app weekly. Charging from the first booking, since willingness to pay is itself a test.
- Cost structure
- Two founders unpaid, hosting and tooling around 300 pounds a month, vetting checks at 40 pounds per walker. Break-even at roughly 120 standing weekly slots.
- Key metrics
- Focus metric: standing weekly slots retained after four weeks. Supporting: enquiry-to-first-walk conversion and walker fill rate. Downloads explicitly excluded as vanity.
- Unfair advantage
- Left blank, honestly. A dense local walker network with repeat relationships could become one, but on day one nothing here would stop a copycat.
The read. Ranking the risks shows the killer assumption is on the supply side: whether enough vetted walkers will commit to fixed midday slots at this commission. The owner-side problem is well evidenced, so the founders' first test is walker interviews and a hand-run pilot of ten standing slots, before any app exists. The canvas took an evening; it redirected the next three months.
Pitfalls
- Filling in the Solution box first and bending every other box to justify it. The canvas reads left to right from Problem for a reason.
- Writing one canvas for several different segments. Where segments differ in problem or channel, Maurya's advice is one canvas each; a blended canvas averages its way to fiction.
- Treating the canvas as a pitch artefact rather than a hypothesis log. Polishing the words is not testing the claims.
- Padding the Unfair advantage box with 'first-mover advantage' or 'our passion'. An honest blank is more useful and more common.
- Never versioning it. The canvas only earns its keep when tests change what is written on it, so date each revision and keep the history.
What the critics say
Scholarship on lean startup methods argues that canvas-driven experimentation gives founders little guidance on where good hypotheses come from, and that testing box-level assumptions one at a time tends towards incremental, local-maximum business models rather than novel ones.
Felin, T., Gambardella, A., Stern, S. and Zenger, T. (2020) 'Lean startup and the business model: Experimentation revisited', Long Range Planning, 53(4).
Critical assessments of the canvas family note the omission of competition and the external environment: nothing on a Lean Canvas forces contact with industry structure, regulation or rivals, so a model can look coherent on the page and be structurally doomed.
Coes, B. (2014) Critically Assessing the Strengths and Limitations of the Business Model Canvas. Master's thesis, University of Twente.
The substitution of the four infrastructure boxes is itself contested. Maurya concedes Key partners was 'the hardest one to remove', and for partner-dependent or capital-intensive businesses the removal deletes the model's most important risks from the page.
Maurya, A. 'Why Lean Canvas vs Business Model Canvas?', LEANFoundry / Love the Problem blog.
Work it through
Fill the nine boxes for one customer segment, starting with Problem and Customer segments rather than Solution, and leave blank what you cannot honestly answer. Your entries persist for this browser session and can be copied out as Markdown or printed.
0 of 9 blocks filled
Sources and further reading
- Maurya, A. (2012) Running Lean: Iterate from Plan A to a Plan That Works. 2nd edn. Sebastopol, CA: O'Reilly Media.
- Maurya, A. 'Why Lean Canvas vs Business Model Canvas?', LEANFoundry. ↗
- Osterwalder, A. and Pigneur, Y. (2010) Business Model Generation. Hoboken, NJ: Wiley.
- Felin, T., Gambardella, A., Stern, S. and Zenger, T. (2020) 'Lean startup and the business model: Experimentation revisited', Long Range Planning, 53(4). ↗