Engineering’s V-Model Method

While the V-model varies across engineering applications, they share a common core. The process starts at the top left, works down the left-hand side of the V, and then moves up the right-hand side. The left side innovates, tests, and validates on paper the best possible solution. The right side builds, tests, and validates the design in the quickest, least costly way. The method captures engineering’s mantra, “build the right thing, and build the thing right.”

Adapted from Kim R. Fowler, Chapter 1 – Introduction to Good Development, Editor(s): Kim R. Fowler, Craig L. Silver, Developing and Managing Embedded Systems and Products, Newnes, 2015, Pages 1-38.

Build the Right Thing

Engineers design new systems “top-down.” After defining the systems’ core requirements, they first come up with a high-level concept. It defines how a system works at its most elemental level. They then work down, eventually reaching the design of individual components. They do so because it leads to designs that are elegant and simple. By design, they don’t mean its aesthetics, but how the thing fundamentally works. 

Elegant, simple designs are more reliable, because there are fewer things to fail and go wrong. They also cost less, as there’s less that goes into making it, and less to manage, repair, and replace once operational. It’s why “simple design” was Steve Job’s famous obsession. 

The keys to coming up with an elegant design are in the first two steps.  

First, it requires precisely defining core system requirements—the most fundamental things the system has to do. It’s about being as clear as possible about the questions to be answered. It’s why step one of Elon Musk’s algorithm is “questioning every requirement” using first principles thinking. The goal is to probe the underlying physics of what the system has to do so that you get to the crux of the problems. 

Second, it requires solving holistically for the high-level concept—coming up with a design taking all of the core system requirements into consideration. The goal is to arrive at an integrated design, not individual solutions to each, or “kludges,” as Space X launch director John Muratore calls them (7). Solving holistically for the high-level concept creates big design synergies that “delete lots of parts”—step two of Musk’s algorithm. 

The Apollo engineers took 14 months to come up with the high-level concept that most elegantly got the astronauts out of the earth’s gravitational pull, into the moon’s orbit, onto the moon’s surface, and back to Earth within the 7-year time constraint. The concept was called “Lunar Orbit Rendezvous” (LOR). It married existing rocket capability with two separate craft—one to land on the moon, and a mother ship that circled the moon and returned home (8). 

The LOR concept created huge synergies. Landing a small rover on the moon’s uncertain surface simplified the challenge, while slashing the amount of fuel necessary. The payload savings reduced the size and weight of the tanks, which reduced the size of the engines, which eliminated the need to innovate new, more powerful rockets. The rover also acted as a life support back-up for the mother ship.

There’s an added benefit of starting with the high-level concept: it establishes performance parameters that cascade down to components. That allows components to be optimally designed for the system. It’s step three of Musk’s algorithm. 

Models and simulations are used continuously to stress test the system’s performance under a range of assumptions. The end goal is a detailed design that credibly works across that range, and has a margin of safety for surprises. 

(7) Muratore, John F. “The Art of Systems Engineering.” Lecture, University of Tennessee Space Institute, October 16, 2008. Accessed at: https://dokumen.tips/documents/4-the-art-of-systems-engineering-rev-1-john-muratore.html

(8) NASA Langley Research Center Office of Public Affairs. “The Rendezvous That Was Almost Missed: Lunar Orbit Rendezvous and the Apollo Program [Fact Sheet].” December, 1992. Accessed at: https://www.nasa.gov/centers/langley/news/factsheets/Rendezvous.html

Build the Thing Right

Building and testing the design is done “bottom-up”: it starts with individual components (e.g., the heat shield on Apollo’s command module), moves to groups of components (e.g., Apollo’s command module), and ultimately the whole system (e.g., the Apollo LOR spacecraft). It’s done this way because the interdependence of a system’s components creates two problems. 

First, it’s difficult to determine why a system isn’t working once operational. When India’s Augmented Satellite Launch Vehicle crashed on its first voyage in the 1980’s, a Failure Analysis Committee explored 37 potential explanations but never pinpointed the cause. 

Anyone who’s run a startup can relate. Is customer conversion low because marketing is bad, the sales force ineffective, or the product too cumbersome or expensive? The result is that it takes more time and money isolating the problem.

Second, fixing the problem part creates ripple effects that require changing other parts, making it more costly. In a software development study, the cost of fixing a problem once the software was operating was 50 times higher than fixing it at the design stage (9). The magnitude of increase is likely higher with hardware.  

In NASA’s case, while the Apollo 11 spacecraft consisted of three main sections, they contained six million parts. It took years of testing at the level of parts and subassemblies before a staged series of unmanned and then manned test flights were conducted, each one taking a step closer to approximating a moon landing. 

By the time Apollo 11 launched from Cape Kennedy on July 16, 1969, an enormous amount of risk had been eliminated.

(9)  Lyon, D. (2012). Systems Engineering: Required for Cost-Effective Development of Secure Products. Global Information Assurance Certification Paper, The SANS Institute.

Key Principles

IVE inverts two closely-held, but problematic, principles of lean startup: that the best way to tackle uncertainty is to use lean and agile processes that rapidly explore lots of potential directions, and that the only way to really learn is to build things and experiment.

Inversion 1: Robust Solutions First, Lean and Agile Processes Second

There’s little value to being lean and agile if a startup’s core business archetype doesn’t have a path to profitability, and when there’s little clarity about performance parameters. In that case, moving fast—or even slow—is simply breaking things. It’s neither an effective nor cheap way of addressing uncertainty.

The core logic of IVE is that you don’t need to eliminate uncertainty—you need to withstand it. The way to do that is by designing a startup so that it’s profitable under pessimistic operational assumptions, and has an added margin of safety to absorb unanticipated costs. It’s how the late Charlie Munger famously thought about investing. 

For example, in the design stage of a London-based legal tech startup founded by the second author, the savings generated for customers meant a price point as high as £750 was credible, but could cause purchase hesitancy, while a price of £400 would be an easy sale. Simulations of unit costs using best-case/worst-case ranges for key operational variables and an annual return on equity of 25% yielded a required mean price of £270, with a best case required price of £160 and worst-case of £415. It was a robust starting point.  

When robustness is the objective, the real work of venture building changes radically. It requires developing a rigorous, first-principles understanding of what makes a business tick. It also requires thinking in confidence intervals, and using simulations to understand how the parts fit together. Once this type of plan is in place, entrepreneurs can use lean and agile processes effectively to prototype and test things out. 

Inversion 2: Build and Experiment to Optimize a Business Model, Not to Discover One

Trying to discover a profitable business model by building and testing different product features, customer acquisition strategies, and price points feels like progress. The fact is, it’s a costly illusion.  

Building and testing stuff is generally an expensive and time-consuming way to learn, even for digital products. Software engineers don’t come cheap.

What makes it particularly bad in lean startup’s case is that, when you don’t know how all the parts in a business model have to work together, you’re left with vague performance parameters to test—e.g., “customers will respond to video advertising showcasing the product.” There’s no way to know if a solution is under-engineered or over-engineered, or when it’s time to change directions entirely.    

A global food company who adhered to lean startup principles came to us frustrated and at a loss of what to try next, having spent 3-years and $12 million testing various product versions, price points, sales channels, and marketing messages for a novel fortified food product, and still falling short of profitability.   

Building and testing has an important place in IVE. But it’s used sparingly and only when the learning objective is to choose among options capable of performing at the level needed for profitability—not to figure out if something simply works. 

For example, the mock-ups below represented alternative designs for one of the legal-tech venture’s products called the Recommended Settlement Report. Both met design requirements specified by the core business archetype. Tests with target customers revealed Option 1 better persuaded customers to make a settlement offer. 

Building and testing in this way is predicated, though, on having clear performance and design parameters for components and key activities. Which brings us to the core business archetype.

Key Concepts

No concept is more critical to understanding IVE than that of the “core business archetype,” or CBA. There’s nothing like it in management science today and creating the right one can make a founder’s life easy or very hard.  

To understand what a CBA is, picture an established market, like the mass market home computer. 

Every competitor today in the home computer market—from Apple, HP, Lenovo, Asus, and others—has a different business model. They have different computer designs and features. They position their brands differently. They use different suppliers and locate assembly plants in different parts of the world. They use different marketing strategies and mediums. Some have financing arms. They retail through different channels. 

Despite these differences, they share a core pattern: let’s call it, “out-of-the-box-ready computing device for all members of the family made of standardized components that runs software applications with graphical user interfaces supported by an open operating system.” While difficult to parse out financials for home computers, given that manufacturers today sell a range of other devices, the average gross margins and operating margins of computer hardware manufacturers over a five-year period hover around 38% and 16% respectively. 

This shared “product/operational model/cost structure” is the CBA pioneered by Steve Wozniak, Steve Jobs, and others from the Homebrew Computer Club of the 1970s. It’s the integrated solution that established basic commercial viability for the computing functionality. Every home computer company since then uses it. 

It's worth emphasizing: the CBA does its work on a functionality, not a product idea. It solves for essential, core business requirements, like driving customer adoption, blocking competition from entering, and controlling costs. It ends up defining the high-level shape of the product and, operational model and, in so doing, defines the average cost curve for the industry. 

There’s not enough detail in the CBA to start the business. That’s what each company’s business model provides. The business model is a company’s strategy for deploying the CBA so as to outcompete others. It details out the product and operational model, and determines where each company lands on the average cost curve. It governs each company’s profitability. 

Designed well, the CBA shapes the customer solution in such a way that it also eliminates the biggest roadblocks to performing core business requirements. In so doing, it shifts the average cost curve down, and reduces the burden on firms’ business models, eliminating otherwise costly components. 

Take the case of the home computer CBA. The CBA eliminated the otherwise intense marketing challenge of convincing everyday people to learn a computer language and solder together components, as hobbyist computers at the time required. Its open architecture, which allowed partners to develop applications from arcade-style games to mortgage cost calculators, eliminated the enormous amount of working capital that would have been required to build a critical mass of applications for all family members. 

The CBA also eliminated customer service and technical support, as it could rely on family members comfortable with the technology to help others. That same dynamic could be leveraged to drive customer acquisition, the way breakfast cereal companies harness the nagging child to encourage parents to buy, and to keep competitors from stealing customers, as switching to a different brand computer required convincing several family members. The use of standardized components eliminated scaling barriers, including inventory carrying costs and sourcing complexity. 

Conversely, if the CBA is badly designed, even the most optimized components won’t make the business model profitable.

Blue Apron Case Study

Consider Blue Apron—a pioneer of the meal-kit subscription company that launched in 2012 and went public in 2017. 

To attract customers, they expanded to 50 meal options; partnered with celebrity chefs; launched ready-to eat and low-cost options; created Weight Watchers approved plans; and formed sales and marketing partnerships with Blue Cross Blue Shield, Planet Fitness, Jet.com, Walmart Marketplace, and Amazon. To cut costs, they automated distribution centers; launched an order management system across its warehouses; sourced ingredients directly from farmers and ranchers, and passed shipping costs onto customers. 

The company, however, remained unprofitable, suffering a net loss of $80 million in 2022 with accumulated losses reaching $700 million. In June 2023, Blue Apron sold its operational assets to packaged meal provider, FreshRealm.  

This isn’t a Blue Apron problem. Aside from a temporary Covid lockdown blip, every subscription meal kit company has struggled with sustained profitability, including Hello Fresh, the market leader with sales of $7 billion and 60% market share in the US. The company is pivoting into ready-to-eat meals, quick lunches, human-grade pet food, and premium fresh beef and seafood, with the goal of becoming a “fully integrated food solutions group.”

Since a well-designed CBA works by eliminating things from the business model, the likelihood of ending up with a robust CBA by filling in components on a business model canvas is extremely low. To give the business model the best chance of success, the CBA needs to be solved for directly. Which takes us to IVE’s V-model.

Key Steps: 

IVE’s V-model works the same basic way as engineering’s V-model. The left-hand side flows top-down to solve for the CBA on paper, such that the solution is robust and has a margin of safety. The right-hand side flows bottom-up. After defining the optimal design of components, components are prototyped and tested in isolation. They’re then brought together and tested in a functional group, like customer acquisition. It culminates with a pilot test of a single business unit. (Please note that the application of the right-hand side of the V model is not in scope for the content of this course).

Build the Right Startup

The left-hand side typically takes six months to complete. It starts off with the functionality of interest, and research is done to determine the prime use case—the broad use case with the greatest potential. In our experience, most founders quickly jump to a basic product concept or technology application. In those case, the idea is first stripped-back to its essential functionality. 

For example, in the case of a tech startup we advised whose technology was a high-altitude kite that generated electricity, we defined the core functionality as “70 kilowatts of energy generated through a truck-sized airborne kite flown at high altitude.” The prime use case was found to be “organizations conducting temporary operations in supply-chain constrained areas”—like natural resource extraction, military operations, and aid work.

The CBA is designed in three stages, its commercial potential expanding with each. Each stage focuses on a sequence of roadblocks to core business requirements applicable to every startup, regardless of industry. They are first principles of startup performance. 

We’ve put them into a tool called the Core Business Archetype Guide. In addition to sequencing the 13 roadblocks, the guide also contains key research inputs needed to define and evaluate an elimination strategy. Summaries are shown below.

Stage one is about basic commercial viability. It focuses on the six most fundamental roadblocks to getting customers to pay a price for a solution that’s higher than what it costs the company to deliver—what we call creating, exchanging, and retaining surplus value. Eliminating these six roadblocks blazes the initial path to profitability and establishes the essential CBA.

For example, Ant Financial Services, the financial arm of Chinese e-commerce platform Alibaba, eliminated the cost bottleneck roadblock that made it impossible for banks to cost-effectively lend to small and medium enterprises lacking formal financial reporting. By applying machine learning to the transaction data generated by businesses on the platform, Ant could quickly tell how well a business was doing, and extend relatively small credit sums at the right interest rate (10).

Elon Musk’s Space X is trying to crack these fundamental roadblocks now, as the cost of today’s single-use rocket boosters makes it impossible to make space travel affordable for the average person. Space X is pursuing multiple strategies, including re-using rocket boosters and harnessing the spacecraft to generate additional revenue by deploying Starlink communications satellites. 

Stage two eliminates four roadblocks that would otherwise hamper selling the essential CBA. They arise from the unique adoption challenge that new-to-world solutions pose—i.e., the routines aren’t already part of or “normalized” in people’s everyday lives, nor is it already “monetized” and part of their budgets.

Fintech company eToro, for example, is built around a social investing strategy that lets clients view and copy other clients’ trading portfolios. It eliminated two roadblocks—systematic and idiosyncratic product risk—that caused people intimidated by numbers and finance to shy away from investing products.

Solar energy providers have made significant inroads into the residential home energy market through a “pay-as-you-save” pricing and payment strategy initially pioneered by entrepreneur Jigar Shah. It eliminated the value perception gap and value timing gap roadblocks by having customers pay for the energy they consume via the monthly bill from their local energy provider, rather than buying solar panels. 

Eliminating these roadblocks results in an enhanced CBA. It widens the path to profitability by lowering the burden ultimately placed on branding and marketing to acquire customers.

Stage three eliminates three final roadblocks that would otherwise hamper the enhanced CBA’s ability to scale. The focus here is on identifying and eliminating business operations that would become complex, costly, and hard to manage as the company grows.  

A&W Rootbeer, the very first fast-food restaurant founded in 1924, propelled the industry by pioneering the use of franchising as a strategy for eliminating the capital-intensive activity of setting up a new restaurant—a product activity bottleneck. GiffGaff, a UK-based mobile virtual network operator aimed at millennials that is shaking up the industry, eliminated the information-activity bottleneck of technical support through a community of customers who receive rewards for addressing queries on its online customer service forum.

To identify these bottleneck activities, the enhanced CBA is mapped. The map traces the flows of product, information, and money down to and back from the customer. Based on the assumptions in the CBA, the activities, people, and resources needed to support each of the flows are modelled, and simulations are run to surface high-volume, high-impact activities. 

The 13 elimination strategies and CBA undergo a number of iterations using the simulation to maximize synergies and the margin of safety. At this point it becomes an CBA—a maximum viable core business archetype. 

Build the Startup Right  

The right-hand side turns the CBA into a detailed core business model, and then prototypes, tests, and optimizes it working up from components. It culminates with a pilot test where customers are acquired and product sold. Depending on the nature of the product, this takes between 12 to 24 months. The output is a maximally viable business unit.

What makes this a different exercise from how prototyping and testing are done today originates in the step at the bottom of the V-model in between the two sides—decomposing the CBA. 

Decomposition breaks the CBA into three main sections— making and distributing the product, acquiring customers, and getting paid for the product. Those sections are broken into their main assemblies, and so on down until you reach the the lowest-level component, like product features, sales pitches, and customer service call scripts. 

The value of decomposition is that it connects the bottom-most components back to the elimination strategies and performance parameters generated by the simulation. It’s clear how each component needs to work, how well it has to work, and which ones disproportionately impact profitability. With that information, a team knows where to focus, and can research and experiment to determine the optimal component design. 

For example, below we’ve exploded out one complete chain in the decomposition for the legal tech venture. There’s a lot here, even in this one chain. The takeaway is that it shows how the design parameters of one operational component—law students writing a message to a customer— flow down through the decomposition.  

The power of this approach is captured in the words of Ryan Lock, co-founder of a leading digital design firm in London called Planes and partner to the legal tech venture: “We’ve built 15+ ventures over the past 10 years – but the clarity that IVE gave our team of designers and developers was unparalleled. We could move extremely quickly and with confidence given the parameters provided to us.”

Prototypes of components are then built and tested. The goal is to optimize them further. For example, when the winning version of the Recommended Settlement Report for the legal tech venture was prototyped and tested, the design was modified to improve the law student’s efficiency in report writing by providing access to a repository of previous cases.  

This test-to-optimize logic extends throughout the right side. By the time the pilot is launched, there should be confidence that all core business functions can meet key performance parameters. The pilot test itself continues testing and optimizing the core business model, this time under conditions that reflect market conditions as closely as possible and where unknown unknowns can be revealed. 

The pilot is considered a success when the business unit can reliably hit key performance parameters on which profitably depend, like customer conversion rates. The venture is ready to scale.

To be clear, the prospect of startup failure can never be eliminated. For example, in an educational venture we were taking through IVE’s V-model that aimed to teach Chinese children to converse in English, the government made it illegal for private-sector tutors to teach subjects taught in school, like English. By following IVE’s V-model, the risk of going down a dead-end path is greatly reduced, as are the time and cost entailed.  

(10) Zeng, Ming. “Alibaba and the Future of Business,” Harvard Business Review, September-October 2018.

Conclusion

When startup ventures fail, more is lost than the prospect of life-enhancing products. Failure means lay-offs, taxpayer and investor money diverted from other needs, and wasted environmental resources. We owe it to society to invest in creating a startup methodology with the greatest potential for success. 

We acknowledge that IVE has yet to fully prove itself. But we’re confident in the new direction, as we have decades of proof that engineering’s V-model approach works extraordinarily well—and, conversely, that experimenting with parts in the hopes of landing on a new, high-performing system as does lean startup, leads to high-cost, low reliability systems. 

With that in mind, we hope IVE can serve as a launch pad for an invigorated entrepreneurship practice that delivers the reliably profitable markets of tomorrow.