The new product development process in manufacturing: eight stages from idea to launch

What is the new product development process

The new product development process is the structured sequence of activities a manufacturer follows to transform an unmet market opportunity into a commercially available product. It begins well before anyone opens a CAD file and continues well after the first units ship. Every serious manufacturer runs some version of this process, whether or not it is documented and whether or not it is called by that name, because the alternative (ad hoc development) reliably produces slower launches, higher costs, and lower success rates than the structured alternative.

NPD as an end-to-end journey

The NPD process is not a single project with a defined start and end date. It is a recurring capability of the organisation that runs multiple new products through the same disciplined pipeline year after year. Each specific new product may be organised as a project within the NPD process, but the process itself is permanent infrastructure of the manufacturer. That distinction matters because organisations that treat NPD as a series of one-off projects reinvent the wheel each time, while organisations that treat it as a permanent capability accumulate learning across projects, refine their gates and templates over cycles, and improve their success rate over time.

NPD process versus product management versus project management

Three roles get confused in this area, and clearing them up prevents a lot of wasted debate. Product management concerns itself with what happens to a specific product across its lifecycle in the market, from launch through maturity to phase-out or retirement. Project management concerns itself with how a specific initiative gets delivered on time, on budget and to scope. NPD process management concerns itself with how the organisation as a whole systematically creates new products, which stages every new product passes through, and how the governance around those stages works. A single new product touches all three: a project manager delivers it, an NPD process shapes how it gets developed, and a product manager takes over ownership once it launches. The three roles are complementary, not competing.

Why manufacturers need a formal NPD process

Formal NPD processes exist because informal ones produce depressingly consistent failure patterns. Nielsen BASES research found that around 80% of new consumer packaged goods products fail in the marketplace, with only about 30% of the roughly 30,000 new products launched annually achieving commercial success within two years. The same research found that innovations with strong product performance were 15 times more likely to succeed than those with poor product performance, underlining that the difference between success and failure is often not luck but discipline. A formal NPD process does not guarantee success; it eliminates the most repeatable causes of failure by forcing the organisation to validate market need before committing engineering resources, to test concepts with customers before committing tooling investment, and to review each product against its business case at every gate rather than only at launch. Without that discipline, projects drift, sunk costs accumulate, and organisations discover their mistakes only when the product hits the market and does not sell.

How manufacturing NPD differs from software NPD

Much of the widely available NPD literature is written by software product managers for software product managers, and it does not translate cleanly to the manufacturing context. The differences are not stylistic; they are structural, and treating manufacturing NPD as though it were software NPD produces expensive mistakes. Four dimensions separate the two worlds decisively.

Physical constraints: tooling, materials, certification

A physical product requires tooling investments that a software product does not. An injection mould for a plastic housing costs between one hundred thousand and two million euros depending on complexity, and once the steel is cut, changing the geometry means new tooling rather than a software patch. Material selection decisions made in early design determine the cost of goods sold for the product’s entire lifecycle, and switching materials late in development can invalidate months of qualification testing. Regulatory certification for products in medical devices, pharmaceuticals, automotive and aerospace takes months to years and follows documentation trails that must exist from Stage 1 of the NPD process, not be assembled retroactively before launch.

Cost of iteration

Software iteration costs are close to zero. A code change deploys in hours, the incremental cost of shipping the update is effectively nothing, and if the change is wrong it can be rolled back. Hardware iteration has almost no relationship to that cost structure. A new prototype run takes weeks and consumes materials, engineering time and machine capacity. A tooling change costs tens of thousands to hundreds of thousands of euros. A regulatory recertification for a design change can take three to six months. This asymmetry means the manufacturing NPD process has to be far more front-loaded on validation than the software equivalent: get the design closer to right before committing to tooling, because the cost of getting it wrong is orders of magnitude higher.

Regulatory and safety requirements

Software products face regulations mostly around data (GDPR, ISO 27001) and, in some cases, industry-specific compliance. Manufactured products face structural regulatory constraints across their entire lifecycle. Medical devices sit under FDA 510(k) or CE MDR review. Pharmaceuticals sit under FDA or EMA approval processes. Automotive components sit under IATF 16949 and, for safety-critical systems, ISO 26262. Aerospace sits under FAA or EASA certification. Every one of these regimes requires a design history file documenting decisions made throughout the NPD process, and reconstructing that file after the fact is neither possible nor legally acceptable. The NPD process for a regulated product must produce the documentation as it goes, which shapes the templates, artefacts and gate criteria from day one.

Time-to-market horizons

Software MVPs ship in six to twelve weeks in mature product organisations. Manufacturing has no equivalent to that timeline. A functional prototype for a moderately complex product takes three to six months. First production launch takes eighteen to thirty-six months for typical industrial products. Complex products such as automobiles, aircraft or medical devices require three to seven years from concept to launch. These horizons are not inefficiency; they reflect the reality of physical development, and the NPD process has to be designed around them rather than pretending they can be compressed by adopting software methodologies wholesale.

The eight stages of the new product development process

Different sources describe the NPD process in five to eight stages depending on how finely they slice the flow. The eight-stage description below is the version most useful for manufacturing settings, because it separates activities that manufacturers actually organise as distinct pieces of work. A five-stage version combines stages that manufacturers keep separate for good operational reasons.

Stage 1: Opportunity discovery and idea generation

The first stage is the fuzzy front end where the organisation identifies unmet needs and generates candidate ideas to address them. Ideas come from multiple sources: customer research through interviews, ethnographic observation, voice-of-customer sessions and complaint analysis; competitive intelligence through product teardowns, patent scans and market analyst reports; internal R&D through technology roadmaps and blue-sky exploration; sales and service feedback from the field. Techniques used at this stage include design thinking workshops, jobs-to-be-done analysis, and structured ideation sessions. The output is a pool of candidate ideas, typically fifty to two hundred, that will be screened in Stage 2. Skipping this stage or shortening it under time pressure is a false economy: it means later stages will work on ideas that were never properly grounded in real customer need.

Stage 2: Idea screening and concept selection

The screening stage narrows the pool of candidate ideas down to a manageable number of concepts worth developing further. The criteria are usually four: strategic fit with the manufacturer’s direction and portfolio, technical feasibility given current or attainable capabilities, market attractiveness in terms of size and growth, and financial viability in terms of expected returns against expected investment. Scoring models and weighted-criteria evaluation reduce the subjectivity of the decision. The output is a shortlist of three to ten concepts moving into concept development, drawn from the fifty-to-two-hundred pool from Stage 1. The main risk at this stage is eliminating breakthrough ideas prematurely because they look too ambitious against conservative feasibility criteria, so screening frameworks need a category specifically for high-risk, high-reward concepts that would otherwise be filtered out.

Stage 3: Concept development and business case

The third stage develops the shortlisted concepts into detailed proposals with formal business cases. Concept development includes refining the product idea through mock-ups or low-fidelity prototypes, testing the concept with target customers, and iterating based on their feedback. The business case is the more consequential deliverable of this stage: a document that quantifies expected market size, revenue projections over a five-to-seven-year horizon, cost of development, projected cost of goods sold, expected margin, breakeven point, and return on investment. The business case is the document the steering committee returns to at every subsequent gate, so it has to be honest rather than optimistic. Fifty to eighty percent of concepts get killed or sent back for redesign at this gate, and that discipline is exactly what makes the process work.

Stage 4: Product design and engineering

The design stage turns the approved concept into a complete engineering package ready for prototyping. CAD modelling produces detailed geometry. Design for Manufacturing (DFM), Design for Assembly (DFA) and Design for Cost analyses check that the design can actually be produced at the target cost and volume. Material selection commits the product to specific supply chains, cost structures and regulatory implications. The Product Breakdown Structure organises the design into assemblies and components that map to bills of materials. Cross-functional reviews with production, quality, procurement and cost engineering catch problems before they become expensive. The output is a design package complete enough that a prototyping team can build working units from it.

Stage 5: Prototype and validation

Prototyping turns the design package into working units. Early prototypes may use 3D printing, machining or soft tooling to produce alpha units for internal functional testing. Later prototypes use production-representative processes to produce beta units for customer field trials. Validation testing covers functional performance, safety, reliability (typically through accelerated life testing that simulates years of use in weeks), regulatory compliance and manufacturability. Two to five iteration cycles between design and prototype are normal at this stage, with each cycle producing design refinements that get folded back into the CAD models and DFM analyses. At the end of the stage the design freezes, and further changes become expensive because they trigger requalification of tooling, materials and regulatory approvals.

Stage 6: Tooling, industrialisation and pilot production

The sixth stage commits capital to production tooling and validates that the design can be made at scale with acceptable cost and quality. Tooling investment covers injection moulds, dies, fixtures, jigs, test equipment, and any custom machinery required. Manufacturing engineering designs the production line: workstation layout, process flow, quality control points, and takt times. A pilot production run of one hundred to one thousand units simulates real production conditions and surfaces problems that laboratory prototypes could not reveal: assembly ergonomics that slow the line down, tooling that wears faster than expected, quality issues that appear only at production volume. The ramp-up plan defines how the manufacturer will scale from pilot to full production rate, typically over three to twelve months depending on complexity.

Stage 7: Launch and commercialisation

Launch is where the product enters the market. Marketing prepares positioning, pricing, channel strategy and launch communications. Supply chain confirms that component suppliers, logistics providers and warehousing capacity can support the projected volume. Sales teams get trained on the product, its features, its target customers and how it displaces alternatives. Service teams get trained on installation, repair and warranty procedures. Regulatory approvals must be confirmed and documented before launch. The launch itself can be phased (regional pilot followed by national rollout to catch early problems before scaling) or big-bang (simultaneous launch across all markets to capture attention), with phased launches being safer for high-risk products and big-bang launches being appropriate for products where competitive timing matters.

Stage 8: Post-launch review and lifecycle management

The eighth stage begins the moment the product ships and continues through its market life. Formal reviews at 30, 60, 90 and 180 days after launch compare actual performance against the business case: are unit sales matching projection, is customer feedback positive, are warranty claims within expected bounds, is the cost of goods sold tracking to plan. Field data drives continuous improvement in production and, sometimes, product updates or redesigns. Lessons learned from the entire NPD cycle feed a repository that improves the next cycle’s estimates and templates. Decisions taken at this stage include line extensions (variants and derivatives to grow the platform), incremental redesigns (to address quality or cost issues found in the field), or phase-out planning (when the product’s market has moved on).

Three methodologies for managing the NPD process

The eight stages describe what the NPD process does; the methodology describes how it is managed. Three methodologies dominate practice in manufacturing organisations, and they are complementary rather than competing. The Product Development and Management Association’s benchmark research consistently shows that best-performing organisations use structured methodologies, with the top quartile reporting NPD success rates around 76% versus roughly 51% for the rest, and the methodology choice is one of the levers that produces that gap.

Stage-gate, Cooper’s classic model

Stage-gate is the reference framework for managing the NPD process, developed by Robert G. Cooper starting in the 1980s and refined across dozens of research studies since. The model organises development into five to seven stages separated by decision gates. At each gate, the steering committee reviews the deliverables produced in the previous stage against a predefined checklist and makes one of four decisions: go (proceed to the next stage with authorised resources), kill (terminate the project), hold (pause pending resolution of specific issues), or recycle (return to the previous stage for rework). The gatekeepers are typically a cross-functional executive team that owns the portfolio, and the criteria at each gate combine strategic fit, market attractiveness, technical feasibility and financial return. Stage-gate suits regulated manufacturing exceptionally well because its documentation trail supports FDA, EMA and ISO audit requirements naturally.

Agile-Stage-Gate, the hybrid model

Agile-Stage-Gate is Cooper’s own adaptation of stage-gate for fast-moving product environments, formalised in his 2016 work. The outer structure remains stage-gate with its familiar gates and cross-functional governance. Inside each stage, the work happens in agile sprints of two-to-four weeks with iterative customer or stakeholder reviews. The gates become more lightweight (accepting agile artefacts such as demos and sprint results alongside traditional deliverables) but the governance discipline remains. Agile-Stage-Gate suits hardware-plus-software products particularly well, such as Internet of Things devices, wearables and consumer electronics, where the physical parts benefit from stage-gate discipline while the embedded software benefits from agile iteration. Pure hardware development gets less benefit from the agile overlay because hardware iteration cycles are too long for meaningful sprints.

Design Thinking for the fuzzy front end

Design Thinking, developed at IDEO and Stanford’s d.school and popularised in the 1990s and 2000s, is not a substitute for stage-gate but a strengthener of the front end. Its five phases (empathise, define, ideate, prototype, test) focus on human-centered design and customer need discovery. Design Thinking is at its strongest in Stages 1 through 3 of the NPD process, where opportunity discovery, idea screening and concept development benefit from its rigour around customer empathy and rapid concept iteration. It is weaker beyond Stage 3 because tooling, industrialisation and regulatory certification are not human-centered design problems. The combination that works well in mature manufacturing organisations is Design Thinking through the fuzzy front end (Stages 1-3) transitioning into stage-gate discipline from Stage 4 onward.

The NPD portfolio perspective

A serious manufacturer does not run one NPD project at a time. It runs a portfolio of five to thirty parallel NPD projects at various stages, competing for shared engineering resources and executive attention. Deloitte’s 2025 Smart Manufacturing Survey of 600 executives at large US-based manufacturers found that 92% see smart manufacturing as the main driver of competitiveness over the next three years, and a coherent NPD portfolio is one of the practical mechanisms by which manufacturers turn that ambition into results. Without portfolio-level governance, the individual project view misses the trade-offs that determine whether the overall NPD investment produces the intended strategic outcomes.

NPD as portfolio, not as single projects

Portfolio thinking asks a different question than project thinking. Project thinking asks whether a specific NPD project should be authorised on its own merits. Portfolio thinking asks whether the balance of projects across the portfolio reflects the manufacturer’s strategic ambitions. A well-balanced NPD portfolio typically follows something like Cooper’s guideline of twenty percent breakthrough projects (high-risk, high-reward, industry-changing), forty percent platform projects (moderate-risk innovations that establish new product families), and forty percent incremental projects (line extensions and improvements to existing platforms). Portfolios that drift toward all-incremental over-invest in short-term returns at the expense of future position, while portfolios that drift toward all-breakthrough take excessive risk without steady near-term revenue. Only a portfolio view surfaces this drift; the individual project view cannot.

Shared resources across NPD projects

The engineers who make manufacturing NPD work are shared resources by design. A senior industrial designer might contribute to eight active NPD projects. A DFM engineer might participate in twelve. A specialist in a particular material or process might be pulled into any NPD project that touches that specialism. Without a portfolio view of resource load, conflicts appear as project slippages three months later rather than as authorisation questions today. Portfolio-level resource management lets the manufacturer plan hiring, external consulting engagements or outsourcing arrangements months in advance of the moment the constraint would otherwise bite, which is the difference between an NPD portfolio that runs on plan and one that permanently overshoots its schedules.

Governance decisions across the portfolio

Steering committees that govern individual NPD projects one at a time miss the most important decision they should be making: which project to kill to fund a more promising one. Portfolio-level governance forces trade-off conversations because the committee sees all projects side by side, with their scoring, their resource requirements and their strategic contribution. Cross-portfolio scoring and prioritisation surface the projects that no longer earn their place in the mix, and killing a project to release resources for a stronger contender becomes a routine decision rather than a political one. Manufacturers that run NPD as a portfolio rather than as a series of projects report higher kill rates at intermediate gates and, paradoxically, higher launch success rates as a result.

Common pitfalls in the NPD process

Four pitfalls account for most NPD failures that discipline could have prevented. Research from Inez Blackburn at the University of Toronto puts grocery-sector new product failure at 70-80%, and Nielsen data on consumer packaged goods sits at similar levels, but the underlying causes cluster around the four patterns below. Each one is preventable once the organisation names it and builds explicit counter-measures into its NPD governance.

Long development cycles without clear go, kill decisions

Zombie projects are the first pattern: NPD projects that neither move forward decisively nor get killed, drifting through months and years of low-intensity development without ever reaching a definitive gate outcome. The test for zombie projects is straightforward: can anyone in the organisation actually kill this project at the next gate, or is that outcome effectively excluded regardless of what the data shows? Zombies happen when executive sponsors have emotional or political investment in a project and the gate process lacks the authority to override that. The remedy is granting the gate committee real authority to kill projects and holding it accountable for exercising that authority, which requires organisational culture change as much as process change.

Sunk cost fallacy at late stages

Sunk cost thinking is the second pattern: the argument that so much has already been invested that stopping now would be waste, when the honest analysis is that continuing will produce even greater losses. This pattern is most damaging at Stages 6 and 7 when tooling investment has been committed and industrialisation is under way, precisely the moment when the largest remaining investments are still ahead. The discipline that prevents this is enforcing that every gate decision uses a forward-looking business case, evaluating the remaining investment against the remaining expected return, independently of what has already been spent. Sunk costs are historical facts, not decision inputs.

Feature creep and scope inflation

Feature creep is the third pattern: the tendency to add capabilities during development on the argument that since the project is running anyway, one more feature will not hurt. Each addition extends development time, raises cost of goods sold, and adds complexity for the end customer. Aggregated across the development cycle, these additions can double the intended cost, delay launch by quarters, and produce a product too complex for its intended market. The counter-measure is change control with quantifiable business impact: no feature gets added without a documented case for how it changes the business case, and the gate committee has to explicitly approve the change and its impact on schedule and cost.

Weak handover from R&D to production

The fourth pattern is the design-over-the-wall problem: R&D completes a design that meets all its functional specifications but cannot be produced in volume at acceptable cost, and the production organisation inherits a design it cannot make work. The consequences show up as launch delays while production redesigns for manufacturability, quality problems in early production runs, and cost of goods sold above target. The counter-measure is embedding production, quality and cost engineering in the design reviews from Stage 4 onward, so that DFM and DFA constraints shape the design rather than being discovered after design freeze. Ownership of the project transitions from R&D to production only after the pilot production run in Stage 6 confirms acceptable manufacturability, not at any earlier point.

How FlexiProject supports the NPD process

FlexiProject sits in the project portfolio layer of the manufacturing technology stack, above the operational systems (CAD, PLM, MES, ERP) and below the strategic layer where the executive committee sets direction. It does not replace any of those systems; it holds the portfolio of NPD projects together with the phase models, governance workflows and cross-project resource management that the NPD process requires at scale.

NPD portfolio management with phase models

FlexiProject organises NPD projects into a dedicated portfolio with phase-model templates matching the eight stages of the NPD process. Each project inherits a common phase structure with the same gate criteria, deliverable checklists and scoring dimensions, which makes projects comparable at the portfolio level. Individual projects carry their own project charters, business cases, budgets and resource assignments while the portfolio view rolls up total investment, expected returns and portfolio balance across breakthrough, platform and incremental categories.

Stage-gate acceptance workflows

The stage-gate decision points are implemented as acceptance workflows in FlexiProject. At each gate, the steering committee reviews the project against its business case, sees the versioned history of the project charter, and records the go, kill, hold or recycle decision with reasoning attached. The versioned charter and decision archive support regulatory audit requirements from FDA, EMA and ISO regimes, which reconstruct the state of a project at any historical gate rather than only its current state.

Cross-project resource management for shared engineers

Shared engineering resources appear at portfolio level rather than at individual project level. A DFM engineer allocated across eight active NPD projects is visible in one workload view, with conflicts surfacing before authorisation of a ninth project rather than as slipped milestones three months later. Portfolio managers can model the impact of adding, delaying or accelerating specific projects on the overall engineering load and make trade-off decisions with the actual numbers rather than intuition.

What FlexiProject does not do

FlexiProject does not do the engineering work itself. It does not replace CAD for design, PLM for product data management, MES for production execution, or ERP for financial transactions. It does not conduct customer research or capture voice-of-customer inputs (those belong in dedicated market research tools and customer feedback platforms). It does not generate ideas (idea management belongs in dedicated tools such as Ideawake, KaiNexus or HYPE Innovation). It sits in the project portfolio layer, holding the NPD process together across projects, and it integrates with the operational systems around it rather than trying to become them.

Frequently asked questions

How long does a typical manufacturing NPD process take?

Timelines vary enormously by product complexity and regulatory context. A simple product with minimal regulatory oversight (a new packaging design, a variant of an existing product line) can complete the NPD cycle in six to eighteen months. A moderately complex electronic product typically runs eighteen to thirty-six months from concept to launch. Regulated products in medical devices, pharmaceuticals or automotive routinely take three to seven years, driven by certification cycles that cannot be compressed. These horizons are not inefficiency; they reflect the reality of physical development and regulatory oversight, and organisations that promise otherwise typically discover the constraints the hard way.

What is the difference between NPD and product management?

The two disciplines cover different parts of a product’s life. New product development is the process of creating new products, running from opportunity discovery through launch. Product management takes over at launch and manages the product through its market life: pricing, positioning, lifecycle roadmap, feature updates, and eventually phase-out or retirement. In many manufacturers the same team leads a product from NPD into product management, but the disciplines and success measures differ: NPD is measured by successful launches, product management by revenue, margin and customer satisfaction in the installed base.

What is the biggest reason NPD projects fail?

Nielsen and other researchers consistently identify the same primary cause: misreading customer need, or launching products that solve problems customers do not actually have or that they solve less well than existing alternatives. Product quality is not usually the primary problem; product-market fit is. This is why Stage 1 (opportunity discovery) and Stage 3 (concept development and business case) matter so much. Organisations that shortcut those stages to move faster into engineering tend to launch products that work well but do not sell, which is arguably worse than launching products that do not work well because the failure is harder to diagnose.

Do we need special software for the NPD process?

Not for the process itself; the process can be run on templates and disciplined gate meetings. What does need software support is the portfolio of NPD projects that a serious manufacturer runs in parallel. CAD and PLM are needed for the product content itself, and a project portfolio management system is needed to hold the portfolio of NPD projects together with common phase models, gate workflows, cross-project resource management and audit-grade documentation. Trying to run a portfolio of a dozen NPD projects on spreadsheets and email tends to produce exactly the failure patterns described in the pitfalls section.

Can the NPD process be Agile?

Fully Agile in the software sense, rarely, because physical development cycles are too long for meaningful sprint iteration on hardware. Agile-Stage-Gate, increasingly, especially for products that include significant software or firmware components. The pragmatic pattern in mature manufacturers is stage-gate governance for the overall process combined with agile working practices inside individual stages, particularly Stages 3 (concept development) and 4 (design), where iteration is genuinely valuable. Full agile transformation of a manufacturing NPD process is more often marketing rhetoric than operational reality.

The new product development process is the systematic sequence that a manufacturer follows to turn an unmet market opportunity into a commercially available product, running through eight canonical stages from opportunity discovery through post-launch lifecycle management. Manufacturing NPD differs meaningfully from software NPD along four dimensions: physical constraints of tooling and materials, cost of iteration measured in tens of thousands rather than zero, structural regulatory requirements that shape the process from day one, and time-to-market horizons measured in years rather than weeks. Three methodologies dominate practice: stage-gate as the reference framework for manufacturing NPD, Agile-Stage-Gate as the hybrid for hardware-plus-software products, and Design Thinking as the strengthener of the fuzzy front end. The NPD process is a portfolio problem, not a single-project problem, and the manufacturers who consistently outperform peer benchmarks run their NPD as a balanced portfolio of breakthrough, platform and incremental projects with disciplined gate governance and honest cross-project resource management. FlexiProject provides the project portfolio layer that holds an NPD programme together at scale, with phase-model templates, stage-gate acceptance workflows, cross-project resource management, and audit-grade documentation for regulated products. It integrates with the operational systems around it rather than trying to become them. If a manufacturer’s NPD programme has outgrown spreadsheets and email and needs a portfolio system that models the actual eight-stage NPD process, thirty days of full access with no credit card is a practical way to test the fit.