Understanding innovation · Part I · The fundamentals

03 · Generating ideas: creativity, design thinking and selection

12 min de lectura

In brief

Ideas do not come from nothing: research shows that creativity is predominantly the recombination of existing knowledge, and that the most productive sources are identifiable (users, knowledge imported from other domains, those who bear the problem).
Structured ideation methods work differently from the way corporate culture describes them: group brainstorming, in its classic form, produces fewer ideas than the sum of the individuals working alone; the variants that separate individual generation from collective discussion work better.
Design thinking is a useful method when the problem concerns the experience of a user and is ill defined; it has documented limits when the dominant constraint is technological or economic.
In the generation of ideas, quantity precedes quality: the evidence on innovation tournaments shows that the value lies in the tails of the distribution, and the tails are reached through volume.
Selection is as fragile as generation: the documented biases (overconfidence, confirmation, falling in love with the solution) are countered with explicit criteria, alternatives compared against each other, and decisions justified in writing.

Section 1

Where ideas come from

The popular representation of creativity as an individual flash holds up badly against the evidence. Research on the generation of ideas converges on a less romantic and more useful conclusion: new ideas are, for the most part, recombinations of existing knowledge applied to a context that the combination had not yet reached. Hargadon and Sutton's study of the work of IDEO (1997), one of the most cited ethnographies in the discipline, describes the design firm as a knowledge broker: its creative productivity comes from working for different sectors at the same time, storing solutions seen in one domain and proposing them again in another where they turn out to be new. Organised creativity is above all a broad memory plus transport between domains.

This conclusion has an important corollary for anyone managing innovation processes: the sources of ideas are identifiable and can be cultivated, they are not random. Three deserve particular attention.

Users. The research of Eric von Hippel (starting from The Sources of Innovation, 1988) has documented that in many sectors a significant share of innovations does not come from producers but from users, and in particular from lead users: people who experience today the needs that the general market will meet tomorrow, and who derive such a high benefit from the solution that they build it themselves. Scientific instruments, sports equipment, software: the sectors in which the phenomenon has been measured are numerous, and the operational lesson is that looking for whoever has already improvised a homemade solution to the problem is one of the quickest ways to find both confirmation of the problem and the embryo of the solution.

Imported knowledge. The IDEO mechanism: solutions that are mature in one sector are radical innovations in another. The capacity of an organisation to absorb external knowledge, recognise its value and apply it (absorptive capacity, Cohen and Levinthal, 1990) depends on the breadth of its internal knowledge base: you import well only what you are able to understand. It is the structural argument in favour of multidisciplinary teams, which chapter 13 will take up, and of organisations that work on several sectors at the same time.

Those who bear the problem. Chapter 02 established that the problem comes before the idea; here it should be added that the bearer of the problem is also a source of ideas, with a symmetrical merit and defect. The merit: they know the real constraints of the process better than anyone. The defect: they tend to formulate the problem in the terms of the solutions they already know, asking for a better version of what exists. The work of ideation often consists in going back from the request as formulated to the underlying job, and from there reopening the space of solutions.

Section 2

Structured methods, and what the evidence says

Brainstorming and its paradox. The group method codified by Osborn in the 1950s (defer judgement, aim for quantity, build on other people's ideas) is the most widespread and the most studied, and the experimental evidence is largely unfavourable to it in its classic form: the research of Diehl and Stroebe (1987) and later meta-analyses show that groups in session produce fewer ideas, and not better ones, than the same individuals working separately (nominal groups). The causes identified are production blocking (in a group people speak one at a time, and ideas evaporate while waiting for a turn), evaluation apprehension and free riding on the contribution of others. The variants that work better separate the functions: individual generation in writing first, sharing and collective building afterwards (brainwriting and similar formats). The lesson is not that collaboration is useless: it is that generation and discussion are different activities, and mixing them damages the first.

TRIZ: inventiveness as a repertoire. At the opposite extreme from improvisation stands the theory of inventive problem solving developed by Genrich Altshuller from the systematic analysis of tens of thousands of patents. The underlying observation: technical problems, beneath the surface of their specific domain, fall into a limited number of recurring contradictions (improving one parameter by worsening another: stronger but heavier, faster but less precise), and the inventive solutions documented in patents reuse a finite repertoire of principles (segmentation, inversion, preliminary action, and so on up to the forty canonical principles). TRIZ is demanding to learn and its effectiveness depends on the quality of the application, but its premise is consistent with section 1: inventiveness is disciplined recombination, and those who know the repertoire start with an advantage over those who improvise.

Innovation tournaments: volume as a strategy. Terwiesch and Ulrich (Innovation Tournaments, 2009) formalise a statistical point with direct operational consequences: the value of innovation opportunities is distributed in a strongly asymmetrical way, with a few excellent opportunities in a long tail of mediocre ones. If the value lies in the tails, the rational process is a tournament: generate many opportunities at low cost, filter them in successive rounds with progressively more costly and more selective criteria, concentrate resources on the survivors. The design of the tournament (how many ideas at the entrance, how many filters, with which criteria) becomes a measurable management variable. The parallel with the structure of phases and decisions in innovation processes, and with the portfolio logic of chapter 08, is evident: they are the same principle applied at different scales.

Section 3

Design thinking: what it is, when it works, where it stops

Design thinking is the codification, which took place between Stanford's d.school and IDEO, of the working method of designers into a teachable process: empathetic understanding of the user (field observation, interviews), definition of the problem from the user's point of view, broad ideation, rapid low-fidelity prototyping, testing with real users, and iteration. The best known popular case remains the exercise of the shopping trolley redesigned by IDEO in a week, filmed by ABC in 1999: not for the trolley itself, but because it makes the process visible (observation in supermarkets, ideation in volume, working prototype, field test, all in days).

The documented strengths of the method lie in its home ground: problems centred on the experience of a user, ill defined at the outset, where the main risk is solving the wrong problem brilliantly. Structured empathy reduces that risk; rapid prototyping buys information at low cost, fully consistently with the economics of stopping early seen in chapter 02.

The limits are equally documented, and they should be declared because the method has been the object of a promotion that extended its application beyond its domain of validity. First: when the dominant constraint is technological or scientific (does the mechanism work? do physical limits allow it?), empathy with the user does not answer the question; technical verification is needed, and design thinking does not replace it. Second: desirability does not imply economic sustainability; a prototype that is loved in testing may have unit costs or channel dynamics that make it a bad business, and the method does not contain the instruments to notice this. Third: recent academic and professional criticism (collected for example in the debate in MIT Technology Review, 2023) documents ritualised implementations, reduced to sticky-note workshops with no follow-up, in which the method produces innovation theatre rather than innovation. The balanced synthesis: design thinking is an instrument for the desirability part of risk, to be integrated with technical and economic validation, not a total method.

Section 4

Selecting: the biases and their antidotes

Generation produces options; selection decides where the capital goes, and it is as fragile as generation because judgements about ideas are systematically distorted. Three biases are documented with particular robustness in the innovation context. Overconfidence: the calibration between estimated probability and the real frequency of success is poor, and it worsens precisely among the proposers, who are also the best informed. Confirmation seeking: once the hypothesis has been formulated, the subsequent investigation involuntarily looks for the evidence that supports it and discounts the evidence that threatens it; interviews with potential customers, badly conducted, are a confirmation machine. And falling in love with the solution, aggravated by sunk costs: the more that has been invested in an idea, the more stopping it appears to be a waste, whereas past costs are unrecoverable by definition and the only rational question concerns the return on the next euro.

The known antidotes are procedural, not psychological: you do not eliminate the biases, you build processes that limit their effect. Explicit selection criteria, fixed before examining the alternatives, because criteria decided afterwards bend towards the preferred candidate. Alternatives compared: never assess an idea on its own, because comparison forces the dimensions of merit to be made explicit; a choice is justified only if it can say what it was preferred to. Written decisions, with the reasons and the evidence available at the moment of choice, which make honest retrospective learning possible (what did we know, what did we weigh badly). And the separation of roles between those who propose and those who decide, which chapter 07 will find again in the structure of phased processes: the proposer brings the evidence, a decision maker who has no sunk costs in the idea applies the criteria.

The chapter closes with the distinction that forms the bridge to the following ones: a selected idea is still nothing more than a well chosen hypothesis. The passage from idea to concept (the mechanism described, the physical and economic limits verified at least on paper, the risks named) is the first act of development proper, and it belongs to the risk management process that chapters 06 and 07 treat at length.

In the Volcano method

This chapter covers phase F3 of the path (ideation) and explains its declared requirements: compared alternatives and a justified choice are exactly the procedural antidotes of section 4, and their presence as a condition for closing the phase builds the defence against selection biases into the method. The question that opens the phase ("are there plausible routes, and which one deserves the concept?") presupposes the generation in volume of section 2: a single route is not a choice. The passage from F3 to F4 takes up the final distinction of the chapter: the idea is a hypothesis, the concept is the idea converted into a verified mechanism, and theoretical verification is declared to be full science. The sources of ideas in section 1 correspond to the doors of the method: the bearer of the problem (F1), whoever brings an idea (valued at 5 to 10 per cent of the project plus a licence), whoever brings a technology (assessed by the committee and converted into a shareholding); the multi-sector network of independent companies is the structure that makes possible the transport of knowledge between domains described by Hargadon and Sutton, and the multidisciplinary core is the base of absorptive capacity that allows the value of what enters to be recognised.

Readings

Frequently asked questions

If the evidence on classic brainstorming is so unfavourable, why does everyone use it?

Because it performs real social functions (cohesion, a sense of participation, visibility of ideas to the group) and because its apparent cost is low, while the loss of productivity is invisible: nobody sees the ideas that production blocking caused to evaporate. The correction is not to abolish collective work but to sequence it: individual generation in writing first, discussion and common building afterwards. The social function is preserved and the volume is recovered.

Is design thinking obsolete?

No: it is an instrument with a domain of validity, like all of them. It works when the problem concerns the experience of a user and is ill defined at the outset; it does not answer questions of technical feasibility or of economic sustainability, and used alone, or reduced to a workshop ritual, it produces innovation theatre. Integrated with technical and economic validation, it remains the best method for the desirability component of risk.

How many ideas are really needed before choosing?

More than instinct suggests, because the value lies in the tails of the distribution and the tails are reached through volume: the tournament logic calls for many options generated at low cost and progressively more selective and costly filters. The right number depends on the cost of generation and of assessment, but the warning signal is clear: a single route on the table is not a choice, and an ideation phase that produces a single alternative has not done its job.

How do you avoid falling in love with your own idea?

Not through willpower: confirmation bias and sunk costs act even on those who know about them. The antidotes are procedural: selection criteria fixed before examining the candidates, alternatives always compared with each other, decisions justified in writing with the evidence available at the time, and separation between those who propose and those who decide. This is why a serious process treats these requirements as formal conditions for advancing, not as optional good practice.

← Previous chapter: 02 · The problem before the idea Next chapter: 04 · The forms of innovation →

Back to the Introduction.

We choose to go to the Moon in this decade and do the other things, not because they are easy, but because they are hard.
John F. Kennedy, 1962

Tax close before 31 December Turn your tax burden into a return