Innovation is frustratingly hit-or-miss. More than 90% of high-potential ventures fail to meet projected targets, while roughly 75% of the products released each year bomb.1 Few established organizations remain dominant over time, as revitalization efforts fail or backfire, costing companies time and money and creating openings for competitors; even fewer generate above-average shareholder returns for more than a couple of years.

These failures are often attributed to a lack of money, talent, or luck. But we think the underlying cause is that innovation in dynamic environments — those characterized by novelty, resource constraints, and uncertainty — is rife with critical tensions. When left unaddressed or mishandled, these tensions sink teams and organizations. Until now, there has been little focus on these tensions in practice or theory, leaving leaders blind to their existence and without the rigorous approaches needed to successfully manage them.

[Here are the first four questions to ask. The authors’ discussion of #4 is especially insightful.]

1. Should you be flexible or disciplined when capturing growth opportunities?

A small, U.S.-based security software company received a call from a customer prospect in Germany. To capture the business and meet cash demands, the company chose to enter the German market. It subsequently entered additional overseas markets in a similar manner. “It was more like we were drawn in rather than made a conscious decision,” a company executive told us.

Seizing opportunities as they arise is consistent with the conventional wisdom that companies must move quickly in dynamic markets. But there is an underlying tension here. Acting fast leaves less time for deliberation, so companies can easily end up with an incoherent portfolio of mismatched opportunities. A disciplined approach, in contrast, enables strategic alignment and sets a path for future opportunities, but it can come at the cost of some quick wins.

Our research with successful organizations shows how to resolve the tension. During the opportunity selection phase, it’s better to be disciplined — spending time studying prospects and devising a plan to capture the best ones rather than those that are easiest to attain. In this way, an organization can accumulate knowledge and experience, using early opportunities to build a foundation for more strategic ones later on. During the opportunity execution phase, more flexibility leads to greater success. This helps organizations abandon ineffective products and practices and adopt more appropriate ones.

Increased discipline in opportunity selection creates a foundation for increased flexibility during execution. That’s because more discipline in selection usually reduces the need to rationalize faulty choices later, freeing leaders to approach execution in a more open-minded way.2 Conversely, when leaders take opportunities as they arise, they exhibit a strong tendency to defend their past choices and become more rigid in the way they execute opportunities.

A Singapore gaming company offers a good example of being disciplined first and flexible later. The company took its time conducting customer interviews and studying market adoption trends before choosing Japan as the first market to enter in its global expansion. When it executed this opportunity, however, it quickly discovered that its plan to sell digital content to Japanese wireless providers meant going head-to-head with entrenched Japanese companies. Once managers realized this, they changed their plan and instead partnered with the entrenched competitors in Japan to sell their content throughout Asia. Their flexibility yielded far greater results than the original execution plan would have.

2. Is it better to differentiate your offering or borrow ideas from competitors?

In established markets, the essence of strategy is choosing to perform activities differently from the way rivals do. In nascent markets, however, this approach makes little sense. When a market (or a business category) is still forming, leaders often don’t know who their buyers, suppliers, or competitors will be, much less which points of distinctiveness are likely to matter most to customers.

The tension underlying this dilemma is rooted in the choice between developing a well-differentiated offering or borrowing ideas that work from competitors. The trade-offs are straightforward. Borrowing is faster and often cheaper and easier, but it doesn’t result in a unique offering. Going for differentiation sets a new product or service apart, but it is time-consuming and resource intensive, and customer demand is uncertain.

Leaders can resolve this tension by engaging in parallel play, a practice inspired by preschool-age children.3 Here’s how parallel play unfolds in various stages of innovation.

Early on, put aside differentiation. Borrow ideas instead. Young children playing side by side imitate one another and borrow one another’s toys, but they rarely play together or try to outdo one another. A similar dynamic occurred in the early days of the ride-sharing market: When Sidecar switched to letting drivers use their own cars and offered an app that featured electronic payments, GPS navigation, and driver ratings, Zimride (later renamed Lyft) and Uber followed suit.

Next, test relentlessly — and then commit.4 When young children play, they usually explore various projects and then stick with the one that engages them most. Similarly, we found that high-performing organizations don’t just borrow ideas — they test ideas and learn from market feedback. Then they use that learning to develop a lucrative business model for creating and capturing value and spend their scarce resources only on that strategy. Burbn is a good example. When an early version of the app, which enabled users to connect, arrange meetups, and post photos, proved too complicated for users, founder Kevin Systrom investigated what they really wanted.5 What he discovered led to a new business focused solely on easy photo sharing, named Instagram.

Finally, pause, observe, and refine. Often, preschoolers at play pause to reflect on their projects before continuing. Leaders we studied acted similarly by initially specifying basic elements of their business models (a product that customers will find superior to existing solutions) while leaving other elements undefined (such as distribution). Early on, Dropbox committed to providing an easy-to-use product and a free-to-paid tiered model for capturing value.6 But it stopped short of tailoring the offer to consumers, who were Dropbox’s primary users at the time, and building operations around file backup, which was the service’s original and most common use. This robust but undetermined model enabled Dropbox to add additional services, such as file sharing and collaboration, and led to profitable new enterprise customers. By the time it filed to go public in 2018, almost a third of its 11 million subscribers were on a Dropbox Business team plan.7

3. Do you follow what data is telling you, or ignore it?

This is a golden age of data, in which new capabilities driven by data analytics promise to turbocharge companies’ disruptive potential. But some innovation leaders overly defer to data and wind up with a culture in which other legitimate decision-making methods — logic, intuition, and qualitative insights — take a back seat. Other leaders appreciate that pathbreaking innovations are inherently contrarian and that evaluating them requires nuance and interpretation. These innovators sometimes ignore data altogether. Resolving this tension between making data-driven decisions and relying on intuition requires knowing when to take which approach.

Our research suggests that you should lean on data when making incremental improvements to existing innovations for current customers but view it more skeptically when transforming products and services in the face of disruption or when introducing breakthrough offerings. Netflix, renowned for its data-driven decision-making, had one of its biggest hits ever when it ignored the data showing that ’80s nostalgia fared poorly, as did programs featuring kids and actress Winona Ryder, and produced the award-winning series Stranger Things anyway.

Leaders can protect potentially disruptive and new-to-the-world innovations by adopting a discerning orientation toward data and a healthy skepticism about insights derived from data. For instance, while Netflix executives use data to inform their decisions when green-lighting programming, they don’t use it as their sole criterion. “You have to be very cautious not to get caught in the math, because you’ll end up making the same thing over and over again,” said Netflix chief content officer Ted Sarandos. “And the data just tells you what happened in the past. It doesn’t tell you anything that will happen in the future.”8

Such caution ensures that leaders don’t rely on data drawn from existing products in established markets to evaluate unrelated innovations aimed at new markets. When Steve Jobs introduced the Macintosh computer, for instance, he leaned on his theory of technology, not numbers. (In the early 1980s, there was no data suggesting that there was an overwhelming unmet demand for desktop computers.) This also prevents innovations from withering on the vine due to unrealistic performance expectations.

4. When do you seek internal help or external help?

Innovators need other people’s help. Alone, leaders are subject to information-processing limitations. Their information is incomplete, and what they know is affected by cognitive bias. Left unaddressed, these shortcomings can result in poor outcomes for individuals, teams, and organizations. Using crowds — or the people around us — is an effective way to overcome the inherent limitations of individual cognitive processing, particularly in entrepreneurial settings where leaders must make swift decisions to address the key uncertainties associated with innovation. But when should leaders seek the advice of crowds inside the organization, and when should they seek the advice of crowds outside it?

The key to resolving this tension is understanding when and how to source internal and external crowds, using a novel strategic framework we call crowd sequencing. Crowd sequencing consists of three steps.

First, use external crowds to address problem uncertainty. Leaders are constantly bombarded with issues, all of which seem to require attention and resources. But often, leaders have an incomplete or inaccurate picture of what’s going on, such that it is difficult to know whether they are focusing on the right problems. They typically rely on focus groups of knowledgeable consumers to overcome this, but our research shows that an unfocused array of people better helps expose leaders’ unknown unknowns. Tapping into a greater diversity of input by sourcing knowledge from a crowd consisting of many outsiders with diverse backgrounds helps leaders find the right problems.9

Second, use external crowds to address demand uncertainty — that is, to determine whether you’ve found the right solution to a problem. A good way to resolve demand uncertainty is to source knowledge from crowds consisting of extreme customers — people outside the organization who would intensely use your product or service and be most likely to recommend it to others, as well as those who would use it rarely, if at all. The heightened sensitivities of extreme users help leaders recognize and better connect with customer needs, beliefs, and desires.

Honda did this when designing its Ridgeline pickup truck. Engineers questioned two types of extreme users: pickup truck lovers, specifically, people who ran businesses out of them, such as electricians and landscapers; and occasional users, such as people who used their trucks only at weekend tailgate parties. From the truck lovers, the engineers discovered a general dislike for traditional tailgates. These customers wanted a tailgate that swung out and could detach to make cargo loading easier. From the weekend tailgaters, the engineers learned that a built-in ice cooler and an electrical outlet (to plug in a TV or mixer) would come in handy. These ideas helped transform a good pickup truck into one of the most popular midsize trucks in the U.S.

Third, use internal crowds to address supply uncertainty. By supply uncertainty, we mean that even after you’ve figured out what people want, you might not have the knowledge necessary to execute. Executing a new solution typically means solving a series of new problems that call for bits and pieces of know-how from varied sources. Though leaders often turn to their closest associates for ideas and expertise, a better approach is to source knowledge from people inside the organization whom they barely know — individuals Mark Granovetter has characterized as “weak ties.”10 Established teams are liable to revert to the same old approaches. When seeking the kinds of novel solutions that executing a new product or service tends to require, it makes more sense to look beyond your usual networks and tap people in other departments or business units.