Forget ‘fail fast.’ Here’s how to truly master digital innovation

“Fail fast” is a popular mantra at companies exploring digital innovation as they try a variety of experiments to see what sticks. Yet failure rarely offers useful business lessons, according to Jeanne Ross and Nils Fonstad, research scientists at the MIT Center for Information Systems Research. In fact, the pair found that this type of thinking often leads to projects that create little if any actual value for the company.

“Leaders often assume that failures will lead to valuable learning. Our data doesn’t support that assumption,” Ross and Fonstad wrote in a recent research briefing. The reason? Most failures have multiple, interrelated causes. “As a result,” they write, “it is difficult to extract a reliable summary of lessons learned.”

Instead of failing fast, companies should “learn fast” by designing initiatives to ensure learning, instead of hoping that failure leads to insight, Ross and Fonstad write. This type of thinking requires a cultural shift from organizational hierarchy to small, cross-functioning teams, and employees should be encouraged to test hypotheses by asking probing questions and admitting what they don’t know. “The challenge is to be much more purposeful about what you’re doing,” Fonstad said.

Ross and Fonstad are part of a team that has spent years studying leading companies undergoing digital transformation, including DBS Bank, LEGO, Toyota, Royal Philips, Audi, Deutsche Telekom, and Posten Norge (the Norwegian postal service), among other global organizations. Ross is the co-author of “Designed for Digital: How to Architect Your Business for Sustained Success.”

This recent research encountered companies attempting digital innovation without learning what customers want; understanding what they were capable of providing; or figuring out how to combine the two in ways that make business sense, Fonstad said. Those innovation efforts are at risk of ending up as “fondue sets” — nice to have, but adding little if any value to the organization, he said. 

Here’s how to design an organization focused on learning fast:

Articulate and test hypotheses

Companies rushing to innovate often make the mistake of starting with an answer. “They don’t take the time to reexamine their early assumptions,” Fonstad said. “Design thinking is all about stepping back and articulating your assumptions. Really take the time to explore.”

Fonstad and Ross argue that businesses should borrow from the scientific method of formulating hypotheses: intelligent, articulated guesses that are the basis for taking action and assessing outcomes. Hypotheses are the starting point for a scientific inquiry, and have long helped scientists work through uncertainty, Fonstad pointed out. They emerge from underlying assumptions, and articulate how those assumptions are expected to play out in a given context.

7-Eleven Japan, the most popular retailer in the country for more than 30 years, has successfully used hypotheses to guide growth, Ross and Fonstad said. The company allows salesclerks to hypothesize about what to stock in stores each week, using detailed sales reports to monitor their performance. Each week counselors visit stores to talk to the clerks about how accurate their hypotheses were and how to adjust accordingly.

Fonstad and Ross found that testing hypotheses can lead to more thoughtful actions and a better understanding of outcomes, a shift from an action-first mindset.

Create empowered, cross-functional teams

A traditional corporate hierarchy, with a top-down flow of information, is not conducive to learning, according to Ross and Fonstad, because learning fast requires teams to test hypotheses instead of acting on information passed along a hierarchical chain. Upper-level management should focus on overarching hypotheses that guide the company’s activities, and then give teams autonomy to act and learn quickly.

This less-hierarchical structure reduces uncertainty and fondue-set projects because cross-functional teams are empowered to quickly start and stop experiments based on what they’ve learned. “That might include stopping their effort because it’s no longer going to realize the kind of value that’s relevant to the organization,” Fonstad said.

Encourage probing questions

“The purposed of hypothesis testing is to limit risk, and where possible, preempt failure,” Fonstad and Ross write in their paper. “For the hypothesize-test-learn cycle to work, people must get into the habit of asking probing questions.”

Such questions help employees articulate the reasoning behind a hypothesis, according to the researchers. At 7-Eleven Japan, counselors asked salesclerks three questions that helped them explain and evaluate the reasoning behind their hypotheses and allowed them to adjust their hypotheses in real time:

• What did you hypothesize this week (that is, what did you order?)
• How did you do? (how well did the product sell?)
• How will you do better next week? (what did you learn, and how will you put that learning into action?)

7-Eleven Japan employees asked salesclerks to explain their assumptions behind a hypothesis, so dubious reasoning could be identified and explored before it came an expensive mistake. The process has led to high inventory turnover and profitability, according to Ross and Fonstad.

“The probing question is the why,” Fonstad said. “Why is knowing this going to help us? The probing questions unearth the assumptions that we have and ensure that what we’re trying to learn is relevant. Because the other mistake is you end up learning for the sake of learning. Just like you end up failing for the sake of failing.”

Shift reward structures

Learning fast instead of failing fast requires shifting reward structures, Fonstad said. Companies traditionally reward employee activity, even if those actions do not result in measurable results or meeting objectives. If the reward structure doesn’t shift to a learning-fast mentality, employees are likely to go back to activities-based goals. 

To reward fast learning, he said, companies should assess whether teams achieve their objectives, such as learning or creating value from an initiative.“I’ve heard so many times, ‘We are changing our culture, we’ve doubled the number of teams that practice,’ and I have to ask, ‘So what?’” Fonstad said. “How has that helped you realize greater impact or great business value?”

The hypotheses approach fits with business cultures that accept some failure as an inevitable, and even useful, result of doing business, according to Ross and Fonstad. “Empowered, cross-functional teams will surely attempt experiments that fail — and hopefully they learn from those failures,” the researchers wrote. “But effective teams are not designed with the idea that they should fail fast. They are most successful if they are designed to hypothesize, question, and learn.”