Point-based design picks one option early and reworks it. Set-based design spends more on early tests to avoid late rework.
Lean product development applies lean thinking to designing new products: create knowledge early, keep design options open and make work flow smoothly. The focus is on learning faster, not on drawing faster.
Allen Ward set out much of the thinking in his book Lean Product and Process Development. Ward died in 2004, and the first edition was published from his manuscript in 2007. Durward Sobek revised it and joined as co-author for the second edition in 2014. This guide covers the principles, set-based design, the chief engineer system, batch size, A3s and how the approach relates to lean startup.
Key takeaways
Lean product development treats design as a knowledge-creation process, not just a production of drawings.
Set-based concurrent engineering keeps several design options open until data shows which one fits.
Toyota's chief engineer system gives one person overall responsibility for a product's concept and success.
Smaller batches and a steady cadence reduce waiting in design work.
A3 reports and trade-off curves communicate decisions on a page and make learning reusable.
What is lean product development?
Lean product development is a method for creating new products that treats the main output of design as knowledge. The drawings and prototypes are carriers of that knowledge.
Ward drew on his study of Toyota. He argued that lean product and process development differs from lean manufacturing because the work is uncertain and creative. You cannot just remove waste from a fixed procedure, because the procedure itself is being invented. See our lean methodology overview for the wider context.
People also search for 'lean new product development'. The terms describe the same field.
That knowledge is expensive to produce. US businesses performed $722 billion of research and development in 2023, up 4.4% from 2022, according to the National Center for Science and Engineering Statistics.
What does efficiency mean in product development?
Efficiency here means more proven value from fewer wasted cycles, not keeping everyone busy. A designer waiting for feedback or an engineer testing a risky idea may look idle, yet save months of rework. Four kinds of efficiency matter:
Value
Time goes to problems and features customers care about.
Learning
Assumptions are tested before big money is committed.
Flow
Work moves from discovery to launch without long waits or handoffs.
Quality
Defects and late design changes are prevented, not fixed at the end.
What are the principles of lean product development?
The principles center on creating knowledge, seeing value from the customer's view, keeping options open, and keeping work flowing. Different authors list them differently, so treat any list as a summary.
Ward's book describes ideas such as defining value and front-loading the design process to explore alternatives. It also covers a leveled flow of work and a strong chief engineer, and it stresses standardization that supports flexibility, not rigidity.
These ideas overlap with the Toyota Production System and share its respect for people and focus on waste. In development, the waste is often rework, waiting for decisions and unused information.
What is the learning loop in lean product development?
The learning loop is a cycle of forming a question, testing it, and recording what you learned so the next project can reuse it. Recorded knowledge builds up over time.
In a typical loop, a team states a design question and builds a quick test or model. It then reads the result and captures it in a short document. That document is then available to future teams.
This is close to the PDCA cycle, applied to design. Teams that do not record what they learned tend to repeat the same mistakes on the next product.
Testing before committing pays because many ideas fail. In Harvard Business Review, Ron Kohavi and Stefan Thomke report that at Microsoft about one-third of tested ideas prove effective, one-third have neutral results and one-third have negative results.
Making assumptions visible
A task such as 'build onboarding flow' hides assumptions: users understand the setup language, they will connect an account, they trust the product with their data, and the integration can finish in time. Name them before building, and rank them by how likely they are to break the concept.
Write each one so it can be tested. 'Operators can complete setup without expert help' can be tested, and 'the product is easy to use' cannot. In hardware, assumptions often involve material strength, supplier capability, build time or tolerance stack-up.
What is set-based concurrent engineering?
Set-based concurrent engineering is a design approach in which a team explores several options at once and narrows them gradually as evidence arrives. It contrasts with point-based design, where you pick one option early and refine it.
Say a team must choose a housing material for a device. A point-based approach picks one early. A set-based approach tests three, discards those that fail on cost or strength, and chooses from the survivors. The example is illustrative, and the next section works it through with numbers.
Studies of Toyota by Allen Ward, Durward Sobek and colleagues, including a 1999 article in the Sloan Management Review, described this set-based practice. It costs more effort early but reduces expensive late changes.
A set-based example with numbers
These figures are illustrative. The housing must cost under $4.00 per unit, weigh under 120 g and survive a 1.5 m drop test.
Three options start in the set: ABS plastic at $2.10 and 95 g, polycarbonate at $3.20 and 105 g, and aluminum at $5.60 and 140 g.
Review one, in week 2, uses the cost and weight data. Aluminum fails both targets and leaves the set. The team records why, so a later product with a different budget can revisit it.
Review two, in week 5, uses drop tests on simple prototypes. ABS cracks in 4 of 10 drops, and polycarbonate passes all 10. Polycarbonate is chosen, and ABS stays documented as the cheaper choice if a future design adds ribs for strength.
Nothing was decided before the data needed for that decision existed. That is the core of the set-based approach.
Matching the prototype to the question
The best prototype is the least detailed one that can answer the current question. A polished model will not tell you whether a part survives repeated use, and a rough rig will not tell you whether customers like the look.
| Prototype | Examples | What it answers |
|---|---|---|
| Concept | Sketches, storyboards, paper screens, rough models | Do people get the idea, and does the team agree on what it is building? |
| Interactive or simulated | Clickable mockups, scripted demos, concierge tests | Can people use it, and do they see enough value to continue? |
| Technical or functional | A test rig, circuit, algorithm or data pipeline | Can the riskiest part work at all? |
| Pilot or MVP | A few units or a narrow release in a real setting | Does it create value in daily use? |
Concept
Sketches, storyboards, paper screens, rough models
What it answers: Do people get the idea, and does the team agree on what it is building?
Interactive or simulated
Clickable mockups, scripted demos, concierge tests
What it answers: Can people use it, and do they see enough value to continue?
Technical or functional
A test rig, circuit, algorithm or data pipeline
What it answers: Can the riskiest part work at all?
Pilot or MVP
A few units or a narrow release in a real setting
What it answers: Does it create value in daily use?
Be clear about what is real and what is simulated. A convincing mockup can be mistaken for a nearly finished product.
Looks-like, works-like and builds-like prototypes
Hardware teams usually need three kinds of prototype, and mixing them up leads to bad decisions:
Looks-like
Tests form, ergonomics and how customers react to the design. It says little about strength or function.
Works-like
Tests whether the mechanism or electronics do the job. It may look nothing like the final product.
Builds-like
Tests whether the design can be sourced, assembled and checked at volume. Treat early builds as learning events, and record assembly confusion, supplier variation, fixture needs and test failures.
Bring manufacturing, quality, procurement and service people in before design choices are locked. Late changes to tooling or suppliers cost far more than changes to software.
What is the chief engineer system?
The chief engineer system is Toyota's practice of giving one senior engineer overall responsibility for a vehicle's concept, design and commercial success. The chief engineer leads through technical authority, not formal power over staff.
The chief engineer writes the concept, holds the vision for the customer and resolves trade-offs between functions. Specialist departments still own their technical expertise.
The idea is to avoid a product that is a compromise assembled by committee. Outside Toyota, product owners or lead architects play a comparable role, though the match is never exact.
How do cadence, flow and batch size apply to design?
In design work, a regular cadence, a visible flow and small batches reduce the time that ideas sit waiting. Large batches of drawings handed over at once hide problems until late.
A cadence means synchronized, regular events, such as a design review every two weeks. Teams know when decisions get made, so work does not stall waiting for an ad hoc meeting.
Reducing batch size means releasing design information in smaller pieces, early and often. Say a team sends a complete drawing set after eight weeks and finds an error. If they had sent a partial set every week, the error, which sat in the week 2 drawings, would have surfaced in that week's review. Instead, six weeks of later drawings built on it, and the team spends about two weeks reworking them. The numbers are illustrative.
A kanban board can make design flow visible and limit work in progress. Our guide to lean thinking explains the flow principle.
How do A3s and trade-off curves support decisions?
A3 reports and trade-off curves put the reasoning behind a design choice on one page, so teams can discuss and reuse it. The A3 format was developed at Toyota.
An A3 states the background, the current situation, the options and the recommended action on a single sheet. It makes the thinking visible so others can challenge it.
A trade-off curve plots how a design parameter affects a performance measure. For example, it might show how wall thickness changes both weight and strength. Teams use such curves to decide with data, not opinion. In an illustrative curve from past tests, a 1.5 mm wall weighs 80 g and holds 300 N, 2.0 mm weighs 100 g and holds 450 N, 2.5 mm weighs 120 g and holds 520 N, and 3.0 mm weighs 140 g and holds 550 N.
With a requirement of 400 N, the team picks 2.0 mm. Going to 2.5 mm adds 20 g for only 70 N more, because the curve is flattening. Both tools capture knowledge in a form that outlives the project, and the next team can read the answer off the curve instead of testing again.
Five questions for a decision review
A decision review turns a prototype into a decision rather than a show. Agree the criteria, such as customer value, technical risk, cost and time to learn, before anyone looks at the options. Then ask:
- What did we believe? The assumption the test was meant to check.
- What did we test? The prototype, the users and the conditions.
- What did we learn? The result, including what surprised the team.
- What changed? The design, the plan or the priority.
- What is still unclear? The next question, and who will test it.
Each review should end in one of four calls: continue, change direction, pause or stop.
How does lean product development relate to lean startup?
Lean product development and lean startup both aim to learn quickly and reduce waste, but they start from different places. The first grew from engineering at Toyota. The second grew from testing new businesses under extreme uncertainty.
Lean startup emphasizes validated learning about customers through a minimum viable product. Lean product development emphasizes building technical knowledge through set-based design and reuse.
A team can combine them. Use startup methods to test whether customers want the product, then use set-based design and A3s to develop it. The MVP tools category lists software that supports early testing.
Lean and agile product development together
Agile gives teams short delivery cycles and frequent feedback. Lean decides whether the work should be done at all and where the waste sits across the whole value stream. Without lean, agile teams can ship unwanted features quickly. Without agile, lean strategy can stay disconnected from delivery.
A combined approach uses lean discovery to find problems and risky assumptions, prototypes to test options, agile cycles to deliver proven increments and value stream reviews to improve the process itself.
Traps that waste product development effort
- 1
Rushing into delivery. Building early feels productive, but once budgets and reputations attach to a solution, it gets hard to act on bad evidence.
- 2
Studying without deciding. Each new insight starts another round of analysis. Learning should lead to a decision.
- 3
Prototypes as sales tools. Polish used to win approval hides uncertainty instead of exposing it.
- 4
Copying a method unchanged. Hardware, regulated products and internal tools each need their own validation steps and release patterns.
- 5
Doing more with less. Reading lean as cutting resources creates hidden queues and quality problems. Remove low-value work to make time for thinking.
Measures that show real progress
Counting features shipped or prototypes built says little. Better measures mix customer evidence, flow and decision quality:
Learning speed
How quickly key assumptions are tested and decided.
Flow
How long work waits between steps, and where.
Rework
How often work comes back because a requirement or fit was misunderstood.
Customer signal
How target users respond to prototypes, pilots and releases.
Quality signal
Where defects and usability problems are found, and how late.
Portfolio focus
How much active work lines up with clear priorities.
A ten-point check for product teams
Use this at the start of a project or in a review. Each line is a question the team should be able to answer yes to:
- Problem clarity: The customer problem is stated in plain, observable terms.
- Customer evidence: There is enough evidence to justify the opportunity.
- Assumption map: The riskiest assumptions are visible and ranked.
- Prototype purpose: Every prototype has a clear learning goal.
- Cross-functional input: Design, engineering, product and operations weigh in before major commitments.
- Decision points: There are planned moments to continue, change, pause or stop.
- WIP limits: Active work is limited to protect focus.
- Feedback cadence: Customer, technical and process feedback is part of the schedule.
- Learning records: Decisions and lessons are written where the next team will find them.
- Process improvement: The team improves how it works, not only what it builds.
If several are missing, the team may be moving on hope rather than evidence. That does not make the idea bad. It means the learning loops need strengthening.
What lean product development training should cover
Training works when it is tied to a live project, with coaching afterward. A workshop alone rarely changes how a team works. A useful program covers:
- Shared language: Value, waste, flow and knowledge creation.
- Value stream thinking: Seeing delays, handoffs and rework in development.
- Customer discovery: Better problem understanding from real conversations.
- Experiment design: Matching each method to the assumption it tests.
- Prototype planning: For software, hardware or service work.
- Decision reviews: Turning results into calls.
- Retrospectives: Improving the team's own way of working.
Mix roles in the room. Product managers, engineers, quality and operations each see different waste. The lean training guide covers training levels more widely.
How do you start lean product development?
Start small, inside one current project, rather than with a new process for the whole department. An illustrative first 30 days:
Week 1: pick one open design decision and write an A3 for it, with at least two options and the data you would need to choose.
Week 2: set a design review every two weeks with a fixed time, and agree which decisions each review will make.
Week 3: build one trade-off curve from test data the team already has, such as old prototype results, and store it where the next project can find it.
Week 4: put design tasks on a visible board with a limit on work in progress, and note how long items wait for reviews. Compare that waiting time after the next two cycles.
Frequently asked questions
What is lean product and process development?
What is set-based concurrent engineering?
What tools are used in lean product development?
What is the chief engineer system?
How do I start lean product development?
How is lean product development different for hardware?
Choosing MVP Tools?
Read the buying guide: what to look for, mistakes to avoid and questions to ask vendors.
Related guides
- Minimum viable product (MVP): meaning, types and examplesMVP meaning explained: what a minimum viable product is, the main MVP types, how to build one step by step, what to measure, and mistakes to avoid.Read the guide
- Obeya: how a big room helps teams see and decide togetherThe obeya room explained: where it came from, what goes on the walls, how a review runs, how it differs from dashboards and status meetings, how to build a first room, virtual and hybrid setups, and how to tell whether it works.Read the guide
- Lean thinking: the 5 principles and the mindsetLean thinking is a way of creating value with less waste. Learn the five principles, the lean mindset, how it differs from lean tools and how managers apply it.Read the guide
- Lean training: formats, levels and how to build a programLean training teaches people to spot waste and improve work. Compare formats, levels and certification, and how to build an internal lean training program.Read the guide
Sources
Books and papers show their publication year. Web pages show the month we last checked them.
- 1.Allen C. Ward and Durward K. Sobek II, Lean Product and Process Development, 2nd edition (Lean Enterprise Institute, 2014)
- 2.James M. Morgan and Jeffrey K. Liker, The Toyota Product Development System (Productivity Press, 2006)
- 3.Donald G. Reinertsen, The Principles of Product Development Flow (Celeritas Publishing, 2009)
- 4.Durward K. Sobek II, Allen C. Ward and Jeffrey K. Liker, Toyota's Principles of Set-Based Concurrent Engineering (Sloan Management Review 40(2), 1999)
- 5.National Center for Science and Engineering Statistics, Business R&D Performance in the United States Increases to $722 Billion in 2023 (ncses.nsf.gov, September 2025)
- 6.Ron Kohavi and Stefan Thomke, The Surprising Power of Online Experiments (Harvard Business Review, September-October 2017)