Skip to content

Prototyping Strategy: Design Thinking, Uncertainty & Failure

Innovation & New Business Proposal - TUHH Institute of Entrepreneurship & Institute of Innovation Marketing, Hamburg · part of my Technology Management MBA · study notes for revision.


Everything in the course up to this point has been about finding out what is true: who the customer is, what the journey looks like, which attributes create satisfaction, what people will trade off. This chapter changes the question. It asks what to do when the honest answer is that I simply do not know yet, and cannot know by thinking harder. That is the normal condition of a new venture, and prototyping is the discipline built for it.

The framing the session uses is blunt. One of the defining characteristics of the entrepreneurial process, and of most corporate innovation projects, is a high level of uncertainty. You cannot remove it by planning, because the missing information does not exist anywhere yet. It has to be created, by going out and generating it. Reducing uncertainty in a targeted way, which is another way of saying learning what was previously unknown, is exactly what prototyping is for.

The word strategy in the chapter title matters. A prototyping strategy sets the direction of the learning process: which unknowns I am chasing, what attitude I bring to them, and where I choose to start. A prototyping tactic is the next chapter: which artefact to build, how rough it can be, and what to measure. Getting the tactics right on the wrong uncertainty is wasted effort, so the strategy comes first.

1 · Uncertainty is the defining feature, and it has two sources

Section titled “1 · Uncertainty is the defining feature, and it has two sources”

The course text splits the sources of uncertainty into factors that come from outside the venture and factors that sit inside it. The split is practical: external uncertainty is something I have to observe and react to, internal uncertainty is something I can partly resolve through my own choices.

External sources outside the venture, largely beyond my control
  • Technology trends - digitalisation and AI shifting what is possible and what is expected
  • Market dynamics and trade - globalisation, tariffs, and shifting flows of goods and demand
  • Regulation - rules such as emission standards that can open or close a market
  • Black swan events - rare, high-impact shocks such as the Covid-19 pandemic
Internal sources inside the venture or innovation project
  • Capability uncertainty - can we actually build, deliver and operate this
  • Financial uncertainty - will the money last, and do the numbers work at all
  • Organisational uncertainty - do we have the team, the structure and the support to carry it
Two very different kinds of not-knowing. External uncertainty is monitored and hedged; internal uncertainty is largely resolved by deliberate experiments that I design myself.

The reason this taxonomy is worth memorising is that it forces an honest inventory. Most teams can list their technology risks fluently and then go completely silent on organisational risk, which is often the one that actually kills the project.

2 · The point of prototyping is targeted learning

Section titled “2 · The point of prototyping is targeted learning”

The central claim of the session, stated almost as a definition: reducing various uncertainties in a targeted way, that is learning the formerly unknown, is at the heart of prototyping. And more than that, the way in which this learning occurs, and how actively it is pursued and managed, is what decides whether the project makes progress at all. Learning is not a by-product here. It is the deliverable.

The session summarises successful innovation as a learning loop, and the loop is deliberately simple enough to run every week.

Identify and prioritise what you need to learnnot everything is equally uncertain, and not every uncertainty is equally fatal
↓
Articulate the question or assumption you are testingwrite it down as something that could turn out to be false
↓
Design the experimentchoose the prototype and choose the users you will test it with
↓
Run it and record the outcomethe notes say it plainly: take photos, capture what actually happened
↓
Interpret the resultand feed it back into the priority list at the top
Prototyping as an experiment cycle. The two steps teams routinely skip are the first, prioritising what to learn, and the last, interpreting rather than merely collecting.

The session also names the three fits that this learning is aimed at, in order: problem-fit (do we understand the problem we are solving, and does that problem matter), product-problem fit (does the proposed solution actually suit that problem, tested with user personas and minimum viable products), and product-market fit (will it penetrate the market, tested through retention drivers and a growth plan). Attitude towards failure sits underneath all three as the prototyping mindset.

3 · The design perspective: problem space and solution space

Section titled “3 · The design perspective: problem space and solution space”

The perspective the course adopts is the design one, rooted in the empathic exploration of user needs, and it works by running iterative ideation and experimentation cycles in both the problem space and the solution space. That phrase is worth unpacking, because most teams only ever prototype in the second one.

Problem articulationproblem space
→
Problem selectionproblem space
→
Concept generationsolution space
→
Concept selection and developmentsolution space
→
Market introduction and commercialisation
The two spaces over time. Prototyping belongs in both: you can prototype to find out whether the problem is real, not only to find out whether the solution works.

The single sentence the slides attach to that picture is the whole economic argument for the chapter: mismatches are expensive. A concept that is beautifully developed for a problem nobody has is a mismatch discovered late, and late is where the cost sits. Every step to the right on that diagram multiplies the cost of being wrong, which is why cheap learning early is worth so much.

Within each space, design work alternates deliberately between divergent activity, which multiplies options and widens the set of choices, and convergent activity, which funnels the options down, eliminates and decides. Ideation and prototyping form a constant cycle between the two, and each pass through the cycle is narrower and more detailed than the one before.

4 · Design thinking, and what it actually improves

Section titled “4 · Design thinking, and what it actually improves”

The course positions design thinking as a user-experience-driven approach to idea generation and product design, and contrasts it with user-preference-driven approaches that dominate traditional engineering practice, such as quality function deployment.

User-experience-driven (for example design thinking)User-preference-driven (for example QFD)
Driven byThe userTechnology and the market
DataQualitative and subjectiveQuantitative and objective
Working mediumVisualisationCalculation
RhythmIterationSpecification

The interesting part is the experimental evidence the session cites, because it is not a blanket endorsement. In a controlled comparison of design thinking against a traditional innovation approach, idea creativity was measured on four dimensions: novelty (originality and paradigm relatedness), feasibility (acceptability and implementability), relevance (applicability and effectiveness) and specificity (implicational explicitness, completeness and clarity).

Design thinking improved compared with the traditional approach
  • Feasibility - the ideas were more acceptable and more implementable
  • Relevance - they applied better to the actual problem and were more likely to work
  • Specificity - they were clearer, more complete and more explicit about implications
Design thinking did not improve the honest limitation
  • Novelty - no measurable gain in originality or paradigm relatedness
  • So design thinking is not a machine for producing wilder ideas
  • It is a machine for producing ideas that are usable, on target and concrete
The reported effect of applying design thinking on idea generation. Worth remembering precisely, because the common claim in industry is the opposite one, that design thinking mainly boosts novelty.

5 · Failure is the most likely outcome, not the exception

Section titled “5 · Failure is the most likely outcome, not the exception”

Before the session reaches strategy it spends a long stretch establishing a base rate, because the whole argument for tolerating failure collapses if you secretly believe failure is rare.

The startup failure rate depends entirely on how failure is defined, and the class notes give three definitions with three very different numbers:

Definition of failureRate
Liquidating all assets, with investors losing most or all of the money they put in30 to 40 percent
Failing to reach the projected return on investment70 to 80 percent
Declaring a projection and then falling short of it90 to 95 percent

For products the picture is the same: roughly 80 percent of new products fall short of their original expectations and end up classified as failed, disappointing or cancelled. The session states this as a law of market failure, that most new products will fail in the market even if competently executed. Competence is not a defence.

Zoom out to firms and the story repeats. Even the longest-lived company on record, a Japanese construction firm founded in 578, eventually ended its independent existence in 2006. Industry after industry shows the same entry-and-shakeout curve, with the number of firms climbing steeply and then collapsing: US typewriters between 1874 and 1936, US picture tube producers from 1949 to 1970, and the US automobile industry from 1900 to 1960. Leadership does not survive a technology shift either, as typewriters moved from manual to electric to word processors to personal computers, computing moved from mainframes to minicomputers to personal computers, and hard drives shrank from 14 inch to 8 inch to 5.25 inch to 3.5 inch, with a different leader at almost every step.

And it is getting harder, not easier. The class notes cite the finding that public US companies now face roughly a one-in-three chance of failing within five years, compared with about one-in-twenty fifty years ago. Infant mortality among young firms is worse still, in both the US and Germany.

6 · Understanding failure: emotions and cognition

Section titled “6 · Understanding failure: emotions and cognition”

The session splits the study of failure into two halves that need completely different treatment. Emotions are about how failure feels and what that feeling does to the next decision. Cognition is about definitions, perceptions and mechanisms, in other words how we classify an event as a failure at all and what we believe caused it. The course adds a third, conceptual layer, which is how failure is built into the design of the learning process itself.

On the emotional side the session works through well-known personal failures to break the assumption that a failure record is disqualifying: a politician who repeated a school year and lost election after election before becoming prime minister at 62 and later winning a Nobel prize in literature; a band rejected by a recording company on the grounds that guitar music was on the way out; a television reporter fired as unfit for TV who became one of the most recognisable people in broadcasting; an engineer turned down for a job by a large carmaker who started building scooters at home and founded his own company.

The in-class exercise that goes with this is the failure resume. Each participant writes down one or two of their own failures, academic, professional or personal, and then stands up in front of the class and talks about it. The point is not confession. It is that saying a failure out loud in front of peers converts it from something shameful into something examinable, which is the precondition for learning anything from it. The reminder attached to the exercise is that neither success nor failure is ever final.

7 · Cognition: are all failures the same?

Section titled “7 · Cognition: are all failures the same?”

The cognitive half starts with a genuinely hard question: was that actually a failure? The slides put up cases and let the class argue.

  • A large juice brand launched new packaging in January 2009 behind an advertising campaign of about 35 million dollars. Within two months sales fell about 20 percent, costing more than 30 million dollars in revenue, while competitors posted double-digit increases. In February the company announced a return to the old design. Expensive, but the reversal took weeks and the learning was unambiguous.
  • A major development programme whose cumulative net orders and deliveries drifted badly apart, ending up delivering roughly four years late. The mismatch itself is the performance impact.

The class exercise that follows is the one to remember: in groups of two or three, develop a system, mechanism or framework to compare and contrast different kinds of failure. The session then offers two published answers.

The spectrum of reasons for failure runs from blameworthy at one end to praiseworthy at the other:

DevianceInattentionLack of abilityProcess inadequacyProcess complexityTask challengeUncertaintyHypothesis testingExploratory testing

left end = blameworthyright end = praiseworthy

Deviance means somebody knowingly broke a rule. Exploratory testing means somebody ran an experiment in a domain where nobody knew the answer. Both produce a bad outcome, and treating them the same way is the fastest known method of killing experimentation inside a team.

The typology of mistakes sorts failures on two axes instead, the cost of the mistake and its potential benefit:

Brilliant mistakes low cost · high benefit
  • A laboratory error that turns into a discovery; losing a job that opens a better career
  • Cheap to make and highly informative - the quadrant a prototyping strategy tries to live in
Serious mistakes high cost · high benefit
  • Getting divorced; a new venture failing
  • Genuinely costly, but they teach something large; survivable only if you can afford them
Trivial mistakes low cost · low benefit
  • A parking ticket; missing a plane
  • Not worth agonising over, and not worth writing up either
Tragic mistakes high cost · low benefit
  • A fatal car crash; drug addiction
  • Expensive and they teach nothing useful - the quadrant to design out of the process entirely

8 · Failures worth having, and failures not worth having

Section titled “8 · Failures worth having, and failures not worth having”

Putting the spectrum and the typology together gives the practical test I actually want, both in an exam and in a project review.

A failure worth having
  • Cheap - the cost of finding out was small relative to the cost of being wrong later
  • Fast - it happened early, while the concept was still cheap to change
  • Informative - it answered a question that was genuinely open
  • Aimed at a real uncertainty - it tested something that could have killed the project
A failure not worth having
  • Expensive - full build, full launch, full advertising spend before the first real test
  • Late - discovered after commitment, when reversing costs months
  • Uninformative - you still cannot say which assumption was the wrong one
  • Avoidable - caused by inattention, deviance or a known process inadequacy, not by uncertainty
A prototyping strategy is essentially a device for shifting failures from the right-hand column into the left-hand one. Note that the difference is mostly about timing and cost, not about how bad the news is.

This is why the course insists that prototyping strategies must include the possibility of failure in order to maximise learning. An experiment that cannot come out negative was never an experiment, and a project plan with no negative outcomes budgeted into it is a project that will discover its bad news at the most expensive possible moment. That is also why the way we think about, interpret and manage failure, emotionally, cognitively and conceptually, is described as a critical skill for every entrepreneur rather than a personality trait.

9 · Attitude, and the culture that supports it

Section titled “9 · Attitude, and the culture that supports it”

Before any tactic, the session asks for an attitude, and describes what an innovative team or company culture looks like in three concrete respects.

  • How user research is conducted - open-ended, actively looking for surprises rather than confirmation, summarised in the instruction to fall in love with the problem, not the solution.
  • How ideas are treated - strong ideas, loosely held. Commit hard enough to develop an idea properly, stay willing to drop it the moment the evidence says so.
  • How ideas are judged - a user is worth a thousand meetings. Evidence from outside the building outranks internal opinion, however senior.

The session then states the sequence for building a prototyping strategy, and the order is the examinable part.

First, find the balanceenthusiasm for your own project, held together with the humility that you are most likely wrong
↓
Develop a prototyping attitudelet external data guide you, and be explicit about the perspective from which results are judged
↓
Second, identify prototyping starting pointswhere in the project the learning should begin
↓
Only then, deploy prototyping tacticstools, fidelity levels, measurement - the next chapter
Attitude before starting point, starting point before tactics. Reversing the order produces very busy teams building artefacts that answer no question anybody cared about.

10 · Establishing problem-fit: how might we, and abstraction laddering

Section titled “10 · Establishing problem-fit: how might we, and abstraction laddering”

The first prototyping starting point in almost every project is the problem itself. The basis of any prototyping strategy is to define and articulate clearly the problem you are trying to solve, to describe how you arrived at that statement, and to present the supporting evidence for it. Journey maps and personas from the earlier sessions feed directly into the Value Proposition Canvas here.

The class exercise turns pain points into workable problem statements. Each team collects at least five pain points from its past user research, picks a single one, and writes five different how might we statements for it using abstraction laddering.

Abstraction laddering works because a problem statement can sit at many levels of abstraction, from very abstract to very concrete, and you can move deliberately between them:

More abstract - how might we reduce the mess and strain of taking out the trashreached by asking WHY is this important
↑
Initial problem - how might we address liquid spills from trash cans
↓
More concrete - how might we design a leak-proof garbage bagreached by asking HOW might we address this
Ask why to climb, ask how to descend. Climbing too far gives a problem nobody can act on; descending too far quietly smuggles a solution into the problem statement.

The assignment attached to it is a good template for my own project: state the how-might-we statement or pain point, provide prioritised supporting evidence that it is a problem worth working on, using the user research techniques from the earlier sessions, then report out. In Value Proposition Canvas terms, this tests the customer circle, while the solution ideas are the square, and every other team gives feedback on both.

IDEO and the redesign of the movie-going experience for a Peruvian cinema chain (Cineplanet). The case is the course’s worked illustration of human-centred design as a culture, a process and a philosophy.

The culture. Nearly everything about the firm is built to sustain the design process: a very flat structure with senior people seated among junior designers, project spaces arranged to force interaction, and offices decorated with the teams’ own creations. Hiring targets T-shaped people, world class in one discipline and fluent across many others, chosen as much for a demonstrated enjoyment of creating together with other people as for expertise. A culture of helping is scheduled into people’s time, so designers between projects assist colleagues on work that is not their own, and staff are reviewed partly on their contribution to that helping culture. The stated core values are: be optimistic, collaborate, take ownership, embrace ambiguity, talk less do more, learn from failure, and make others successful. Embracing ambiguity is treated as an actual skill, the ability to start from a blank sheet of paper and even relish it.

The process. Three phases, each of which deliberately broadens and then narrows, alternating divergent and convergent thinking. The exploratory phase is intensive research with consumers, industry experts, extreme users whose needs and behaviours are amplified relative to the mainstream, and analogous examples from unrelated sectors, such as studying hotel checkout systems and e-commerce sites when rethinking how a cinema takes payment. The concepting phase turns those insights into hypotheses about solutions and, crucially, brings ideas to life visually so that people can respond to them. The prototyping phase makes concepts tangible enough to be used by real customers in a real venue, whether that is a physical object such as a greeting kiosk or a seating array, or an intangible one such as staff trained to a new script, refined only to the point where roughness stops being a distraction.

The philosophy. The process begins and ends with the human experience. Empathy is the mechanism: the team goes out into the world to gain empathy for somebody else’s existence and to understand how a design might change it for the better. Ideas are held as strong ideas, loosely held, so team members own their concepts but abandon or reshape them when the evidence goes the other way. Early sketches are treated as sacrificial concepts, drawn fast precisely so that nobody becomes emotionally attached, and even rejected ones teach why an approach will not work and can be recombined into new testable ideas.

What the team learns from a customer interview, and how it interviews. The interviews run one and a half to two hours, in a relaxed setting, and above all in context: at home, where people actually consume media on phones, laptops and home cinema setups and can point at their devices and DVDs to show their routine, and inside the cinema itself, because asking someone standing in the queue how they feel about the queue gets a far more authentic answer than asking days later. The team prepares an emergent line of inquiry rather than a questionnaire, for instance whether going to the cinema in Peru is a social or an individual experience, and turns it into exploratory objectives and an outline of exercises rather than questions: reconstructing a daily diary of the past week so that the role of the cinema inside a real life becomes visible, or asking an industry expert to sketch a process on paper so the drawing becomes a shared object to steer the conversation with. What comes out of such an interview and can be used directly in a redesign is the latent material: where the cinema really sits among competing leisure and media options, which moments in the experience are inefficient, irritating or unpleasant, how payment and queueing are actually experienced, and the observation that eventually became the project’s organising theme, that movies are social experiences meant to be shared.

What is notable about how the team interacts is as important as what it asks. The tone is open-ended and unhurried; the group is kept deliberately small so the interviewee is not overwhelmed; the designers sketch on the spot and let the person react to something visible; they watch for unusual or interesting behaviours rather than only recording answers; and judgement is deferred throughout, in line with the firm’s brainstorming rules of deferring judgement, encouraging wild ideas, building on the ideas of others, staying focused, one conversation at a time, being visual and going for quantity. This is explicitly not market research: the team does not aggregate input from a representative sample, it gathers rich stories from average consumers and from the extremes of the distribution and condenses them into insights about desires and motivations. Afterwards the team runs downloading sessions, where each member retells what they saw so that observations and stories become abstract themes spanning several users, and fills in a customer profile form whose fields show exactly what is considered worth keeping: the person’s top need, their favourite concept, what the conversation made the team realise it should consider, a few good quotes, unusual or interesting behaviours noticed, initial thoughts or sketches for this particular user, and the three things the team should still remember about the conversation a month later.

The sting in the tail. The concepts the team converged on were not a set of separate improvements but a system, and adopting them would require the client to shift from a company that emphasised and excelled at operations into one organised around customer needs and desires, changing layouts, training, hiring, IT, services and products. Prototyping surfaces organisational uncertainty just as reliably as it surfaces customer uncertainty.

  1. Write the uncertainty inventory. List every external uncertainty (technology, market and trade, regulation, shock events) and every internal one (capability, financial, organisational) that could plausibly stop your project. Force at least one entry per internal category, because these are the ones teams skip.

  2. Turn each uncertainty into a falsifiable assumption. Rewrite each line as a sentence that could turn out to be false, for example that a named segment will pay a named price for a named outcome. If it cannot be false, it is not testable and does not belong on the list.

  3. Prioritise by two questions: how uncertain, and how fatal. The riskiest assumptions are the ones you know least about and that would end the project if they were wrong. Those go first. This is the prioritisation step of the learning loop.

  4. Fix the problem statement before the solution. Collect at least five pain points from your user research, choose one, and use abstraction laddering to write five how-might-we statements at different levels, asking why to go up and how to go down. Keep the level that is broad enough to allow several solutions and narrow enough to act on.

  5. Assemble the evidence that this problem is worth solving, describe how you arrived at the statement, and prioritise the evidence. This is the circle of your Value Proposition Canvas, and it should be tested before the square.

  6. Declare your prototyping attitude in writing. State what result would make you abandon or pivot the idea, and from whose perspective results will be judged. Enthusiasm, plus the working assumption that you are most likely wrong.

  7. Choose your prototyping starting points. For each top assumption, decide whether the open question sits in the problem space or the solution space, and pick the earliest point where a cheap test can produce a real answer.

  8. Design each test so that failure is allowed and cheap. Before running it, write down which result would count as a negative outcome. Aim every test at the brilliant-mistake quadrant: low cost, high potential learning.

  9. Set up the capture and the review. Record outcomes as you go, photos included, then interpret in a session where the team retells what it saw rather than reporting polished conclusions. Separate praiseworthy failures from avoidable ones so that neither gets punished by mistake.

  10. Feed the result into the pitch. The evidence of progress, the risks and the anticipated solution sections of a pitch are exactly the output of this loop, alongside problem and context, business model, team and capabilities, and the call to action.

TermWhat it means in plain words
UncertaintyInformation that does not exist yet and cannot be found by thinking harder, only by going out and generating it
External uncertaintyNot-knowing that comes from outside: technology trends, market and trade dynamics, regulation, and rare shock events
Internal uncertaintyNot-knowing inside the venture: whether we can build it, whether the money works, whether the organisation can carry it
Targeted uncertainty reductionDeliberately choosing which unknown to attack next, instead of learning whatever happens to be convenient
Prototyping strategyThe direction of the learning process: attitude, which uncertainties matter, and where the learning starts
Prototyping tacticsThe organisation of the actual activities: which tool, which fidelity level, how results are captured and read
Problem spaceThe half of the work concerned with articulating and selecting the problem, before any concept exists
Solution spaceThe half concerned with generating, selecting and developing concepts to address the chosen problem
Divergence and convergenceWidening the set of options, then funnelling them down; design work cycles between the two repeatedly
Design thinkingA user-experience-driven approach to idea generation and product design: user-led, qualitative, visual and iterative
Failure resumeAn exercise in writing down and openly presenting your own failures, so they become examinable instead of shameful
Spectrum of failure reasonsA scale from blameworthy causes such as deviance and inattention to praiseworthy ones such as hypothesis and exploratory testing
Typology of mistakesA grid sorting failures by cost and potential benefit into tragic, serious, trivial and brilliant mistakes
Sacrificial conceptA rough sketch shared early precisely so it can be discarded without attachment, the earliest form of prototyping
Problem-fitEvidence that the problem is understood, real and worth solving, established before solution testing begins
  1. Name the four external and the three internal sources of uncertainty listed in the course text, and explain why the split matters for how you respond to each.
  2. State in one sentence what the course says is at the heart of prototyping, and say what determines whether the project makes progress.
  3. What does it mean to run iterative ideation and experimentation cycles in both the problem and the solution space, and what is the one-line argument the slides attach to that diagram?
  4. According to the study cited in the session, which dimensions of idea creativity did applying design thinking improve, and which one did it not?
  5. Give the three definitions of startup failure used in the class notes with their rates, and explain why the spread between them matters.
  6. Distinguish a failure worth having from one not worth having, and explain why a prototyping strategy must deliberately allow for failure.

Next: Prototyping Tactics → - what to build, how rough, and what to measure.