March 1, 2019—With spring on the horizon, we look at trends the fashion industry should keep an eye on. Plus, the infrastructure challenges for delivery drones and three questions for Michael Chui, an MGI partner and expert on AI and automation.
|
|
|
After a couple of rough years, fashion execs were cautiously optimistic in 2018. So, what’s on the industry’s runway for the coming months? |
Let’s look at a few of the trends highlighted in The State of Fashion 2019: A year of awakening (a survey of fashion executives written in partnership with the Business of Fashion). |
First, one that may have some staying power: the influence of “woke” consumers on fashion. Younger shoppers are concerned with social and environmental causes, and they back their beliefs with their wallets, favoring brands that align with their values and avoiding those that don’t. |
The views of Gen Z and millennial consumers are critical. Together, these younger consumers represent around $350 billion of spending power in the United States alone (approximately $150 billion spent by Gen Z and $200 billion by millennials). Gen Z will account for 40 percent of global consumers by 2020. And concern over environmental and social issues is not just for the young: some two-thirds of consumers worldwide say they would switch, avoid, or boycott brands based on their stance on controversial issues. |
A related trend is radical transparency. Recent high-profile data breaches at online fashion companies and in other industries have left consumers wondering whether to share information with brands and retailers. They’re also demanding to know much more about issues like where and how items are made, a design’s provenance, and an item’s quality. |
The areas where fashion companies will be most scrutinized include creative integrity, sustainable supply chains, value for money, treatment of workers, data protection, and authenticity. Given heightening concerns, fashion companies should identify potential business practices that could erode consumer trust. |
Finally, there’s fashion on demand, where tech developments and shifting consumer needs have made just-in-time production a trend to watch. A new breed of start-ups has adopted made-to-order production cycles, and mass-market players are following suit. This is all creating new challenges for companies, but it’s good news for your spring wardrobe. |
|
|
|
OFF THE CHARTS |
Will drone delivery get off the ground? |
Start-ups, high-tech giants, and others have already begun investing in the innovative technologies needed to make delivery and transport drones a reality. But building roads in the sky isn’t easy. Click on the interactive below featuring the infrastructure complexities ahead. |
|
|
|
|
INTERVIEW |
Why this CEO won’t be mad if you lose money |
ShopRunner CEO Sam Yagan (also cofounder of OKCupid and former Match Group CEO) talks about data-based decision making and how requiring people to fail fast is one of the most important attributes of an innovative culture. |
“During annual reviews, I request that every executive quantify his or her failures,” he told McKinsey. “In fact, you start to ask them, ‘Did you take enough risk? You had only a $10,000 failure this year. What if you had had a $100,000 failure? What other returns could we have gotten if that bet had gone the other way?’ I think it’s about celebrating and quantifying failure and being willing to talk openly about it.” |
|
|
MORE ON MCKINSEY.COM |
Banking on digital customers in Indonesia | The country’s rapid shift toward digital banking presents opportunities for both incumbents and attackers to attract and keep loyal customers. To do that, banks must adopt digital practices that reflect a customer-centric mind-set. |
Insurance underwriting: From art to science | For commercial P&C insurance, underwriting excellence remains paramount to performance. To modernize, companies must strike a balance between automation and judgment, as well as autonomy and control. |
Saying no to delays and cost blowouts | Major construction projects are, on average, delivered a year behind schedule and 30 percent over budget. Understanding how work flows—through definition, planning, readiness, control, and execution—can remove bottlenecks and boost productivity. |
|
|
|
What types of AI applications could have the most benefit for social good? |
We looked at 160 applications that cover all the United Nations Sustainable Development Goals, and we found that some of those with the greatest potential are what we call “computer-vision applications.” |
They include everything from disaster relief, where you might use satellite imagery to identify passable roads, to public health, where healthcare workers can remotely diagnose skin cancer and other types of diseases. These AI technologies can be deployed for more complex types of data and imagery but also in traditional databases. We’ve seen nonprofits, which typically have their hands on lots of data, apply AI solutions to their old-fashioned databases, with good results—for example, routing solution-oriented people to places of need more quickly and efficiently. |
What are the obstacles that must be overcome to use AI for social good? |
Chief among them is accessibility to the data needed to train these AI systems. In many cases that data exists, it’s just locked away, whether because a commercial organization sees value in selling it or because of bureaucratic inertia. Governments have lots of data that could be valuable for addressing these social-good challenges, but it hasn’t been made available. |
Another obstacle is talent. People with the skills to deploy these technologies are in short supply. Finally, there’s a set of “last mile” challenges—not having the funding or the connectivity or the right number of people in place. You might have superior insights from AI, but unless you can change conditions on the ground, AI-based solutions are not going to be able to move the ball forward in terms of social good. |
How can we make sure that AI doesn’t pick up ‘bad habits’ like bias? |
That’s a real challenge. Bias often gets introduced into these systems not because the software engineer codes up bad rules. Rather, it’s implicit in the training data that’s used to create the systems. |
When we talk about machine learning, it’s a little confusing. The machine doesn’t run off, learn something, and come back. Actually, we train it with data. The problem often is that the data incorporates bias. And then you do get systems that pick up ‘bad habits’—maybe those of people but, more important, the bad habits incorporated in that data. Understand what the data is, and then understand the system that you’ve created and the models you’ve created using those systems. In many cases, it can be helpful to have third-party validation of those models as well. |
|
|
|
|
|