I have repeatedly seen innovation teams face ill-defined questions that could not be resolved through analysis alone. They required judgment: Which opportunity deserves attention? Is this problem important enough to act on? What does the available evidence justify? Should we continue, change direction or stop?
Yet the rules used to answer these questions were often as complicated as the situations themselves. Weighted scorecards, elaborate evaluation models and multi-stage decision systems combined numerous criteria, assumptions and estimated values. The resulting answer might look rigorous, but few people understood how it had been produced, which assumptions drove it or when the same rule should no longer be trusted.
My argument is that complex situations do not necessarily require complex decision rules. Under uncertainty, teams often need one simple heuristic—or a small number of them—that everyone involved can understand and apply consistently. But simplicity alone is not enough. The heuristic must fit the specific decision and the environment in which it is being used.
Most innovation frameworks give teams a sequence of activities.
Select an opportunity. Research users. Define the problem. Generate ideas. Build a prototype. Complete a business model canvas. Run a pilot. Scale what works.
This creates order, but it can conceal the decisions that matter.
A team can complete every activity and still make a weak decision. It can select an innovation candidate with a weighted score built on invented numbers. It can conduct twenty interviews and mistake recurring statements for evidence of demand. It can receive enthusiastic reactions to a prototype and infer willingness to adopt. It can complete every block of a business model canvas without knowing whether the elements can work together in practice.
The problem is not necessarily the absence of a process. It is often a mismatch between the decision being made, the uncertainty surrounding it and the rule used to decide.
Selecting one candidate is different from allocating discovery capacity across five candidates. Deciding whether a problem deserves further research is different from deciding whether to build a solution. Choosing between two solution concepts is different from deciding how much capital can be exposed to an untested business model. Evidence that is sufficient for the next interview is not sufficient for a product build, a commercial launch or a larger investment.
Innovation work is therefore better understood as a sequence of dependent decisions. Each decision requires an explicit search, stopping, judgment and action rule. The aim is not to find one universal innovation heuristic. It is to select the simplest rule that fits the decision environment—and make that rule understandable to everyone whose judgment and commitment it affects.
A phase does not define the decision
Labels such as problem space, solution space are useful, but they describe the object of attention rather than the decision environment.
Inside the problem space, a team may need to:
search for a relevant Job to Be Done;
decide which job performer matters;
judge whether the job is important and underserved;
determine whether further interviews will change the hypothesis;
or decide whether the evidence justifies entering the solution space.
These are not one decision. They have different error costs, different signals and different stopping conditions.
The same applies to the solution space. A team may be generating alternatives, eliminating infeasible concepts, choosing which concept to test, allocating capacity across several tests or deciding whether the observed response is strong enough to continue. Calling all of this “solution validation” hides the decisions that determine what happens next.
Business model work is even more vulnerable to this confusion. A canvas shows relevant elements and relationships, but it does not tell a team which assumption to test first, which negative result should stop the initiative or how much evidence is sufficient before more capital is committed. A complete canvas can still be a complete set of assumptions.
Interested in working together?
Heuristics are not universal shortcuts
In everyday management language, heuristics are often treated as rough shortcuts: useful when time is limited, but inferior to a more complete analysis. Research on fast-and-frugal heuristics offers a more demanding view.
A heuristic can outperform a more complex model when it fits the structure of its environment. Sparse observations, noisy data and uncertain relationships can make elaborate models vulnerable to overfitting. In such conditions, deliberately ignoring information may improve robustness rather than merely save effort. This is the central idea of ecological rationality: the quality of a heuristic depends on the environment in which it is used, not on its sophistication in isolation.[1][2]
Large companies have also developed simple rules specifically for innovation under uncertainty. At Amazon, teams work backward by writing a press release and FAQ before development begins. If they cannot explain which customer would use the proposed offer, why it would be meaningfully better than existing alternatives and why someone would act, the idea is revised or denied development resources.[3] Intuit asks teams to identify the riskiest Leap of Faith Assumption: the critical behavior that must occur for an idea to work but is not yet supported by evidence. The team then builds the minimum experiment required to test it, defines the success threshold in advance, observes actual behavior and decides whether to pivot, persevere or test another assumption.[4] Adobe’s Kickbox applies an Affordable Loss logic: employees receive bounded resources and a structured sequence for developing an idea, testing it with potential users and building organizational support before seeking a larger commitment.[5]
These rules address different innovation decisions. Amazon’s rule tests whether a proposed value proposition is coherent enough to deserve development. Intuit’s rule determines which uncertainty to test next and what evidence should guide the subsequent decision. Adobe’s rule limits the resources exposed while an idea remains highly uncertain. None is universally valid, and the companies do not necessarily describe them as formal heuristics. But each reduces a complex innovation situation to a shared rule for searching, stopping, judging and committing resources.
A usable heuristic is also more than a memorable slogan. It needs explicit rules. The formal literature commonly distinguishes search, stopping and decision rules. For innovation work, it is useful to make the judgment rule explicit as well:
Search: What information, cues or alternatives should we examine, and in what order?
Stop: When should we stop searching?
Judge: How will we interpret the observed evidence?
Decide: What action follows from that judgment?
“Talk to customers,” “build an MVP,” “fail fast” and “follow the evidence” are not complete heuristics. None specifies what to search for, which signal is decisive, when enough has been learned or what commitment the result authorizes.
Interested in working together?
The sequence runs from commitment to commitment
The early innovation journey can be represented as a dependency sequence:
Search boundaries → innovation candidates → problem space → adoption system → critical assumption → solution space → credible test → business model system → first traction
This is not a linear Stage-Gate process. New evidence can send the initiative backward, split one candidate into several, reveal a different job performer or force the business model to change. The sequence describes what must become sufficiently credible before a larger and less reversible commitment is justified.
The central principle is:
Do not assign a heuristic to an innovation phase. Assign it to a decision whose environment makes the heuristic valid.
The practical question is therefore not, “Which innovation method should we follow?” It is:
What decision are we facing now, what kind of uncertainty surrounds it, and which rule is sufficiently reliable to guide the next commitment?
The free section of INNOVATION& is offered freely, and you are welcome to keep reading it that way. If these articles help you make better decisions in your working life, please consider becoming a paid subscriber. Researching and writing them takes time, and paid subscriptions provide the income I need to continue. Your support also helps keep part of this work available to everyone.
Thanks for reading INNOVATION&! This post is public so feel free to share it.
Start paid learning and application section
The Decision-to-Heuristic Map
The following sequence covers the journey from selecting innovation candidates to the first credible signs of traction. It combines published heuristics with deliberately engineered adaptations for innovation work.
That distinction matters. Take-the-Best, Tallying, Satisficing, Elimination by Aspects, fast-and-frugal trees and 1/N have formal or empirical foundations in decision research. Affordable Loss comes from effectuation research. Their use in this sequence is an application, not a claim that every rule has already been experimentally validated across every innovation context. The rules still need to be calibrated against the environment in which an organization uses them.
