It’s a conundrum buyers face every season. What do I buy? How many? What about colors, sizes and related items? What’s the best way to know? How can I do all that within my budget? And most crucially, how confident can I be in the success of my assortment?
For decades, answers to these questions were divined from studying recent historical data and customer surveys. The challenge is that all this information is derived from the past, and the decisions that follow represent educated — but still risky — gambles.
In today’s retail environment, marked by breakneck competition and constantly evolving customer preferences, this hit-or-miss approach is unreliable at best: Look at H&M, reported in The New York Times last March as having some $4.3 billion in unsold items. The company’s predicament represents just a fraction of an industry-wide problem.
Now try to imagine a scenario where concepts can be tested prior to making a financial commitment. Artificial intelligence and machine learning — the cutting edge of the digital revolution — can harness the power of predictive analytics to make that possible.
Detroit-based Shinola has begun using a tool that does just that. The manufacturer and retailer of high-end watches, bicycles, audio products and leather goods operates over 30 Shinola-branded stores throughout the United States and has partnerships with several third-party retailers in Europe; CEO Tom Lewand says MakerSights software represents a true quantum leap.
“Prior to using MakerSights, we simply looked at sales histories and used our best judgment to predict future trends,” he says. “Since we started with MakerSights we’ve tested some 300 products with the system, which generated about 30,000 customer responses. Those have in turn generated almost 9 million product reviews.”
MakerSights leverages AI to provide guidance through each phase of a product’s life: conception, creation, decision (go or no go), sales monitoring and analysis, and end-of-season review. The platform operates on a cyclically repeating closed data loop model. Responses to immersive consumer surveys are gathered, usually by mobile device. The raw information then goes through a “data hygiene” algorithm to assure its quality and accuracy. Finally, the data is synthesized with results of internal product testing to provide real-time actionable takeaways.
“To make its recommendations, the program taps into three major sources (or data points) of information,” says MakerSights co-founder and CEO Dan Leahy. “First, we look at sales data — how things have performed in the past — followed by the current voice of the customer which we statistically quantify, and then conclude with an internal hypothesis, where we consider the creativity and intuition that goes into the product. We believe all three of these serve an important purpose.”
In the end the individual merchant’s decision is ultimately most important, but “the way to go about that is to be armed with knowing what the consumer thinks and what we know internally,” Leahy says.
Since they’re based on real-time data, MakerSights’ recommendations can reliably be used in all phases of a product’s lifecycle. Historical data on consumer wants informs the brand on concept and design, mobile-first interface data from consumers provides guidance to building the line and the resulting hypotheses inform specific buying strategies, such as buy depth and allocations. Once a client begins using MakerSights, the program intuitively observes the accumulated data patterns and uses them in formulating future conclusions.
MakerSights is currently preparing for the release of a new edition of its software, slated for early this year.
Leahy says the new iteration will retain the strengths of the original while taking a more nuanced approach. “The different decision points that happen on the way to market differ between the individual item and the type of organization. Each decision point presents different questions, different constraints and different opportunities. One size doesn’t fit all,” he says.
“For example, at a given point in time, a merchant may not yet need to know how many units it needs to sell, but is looking for other answers: ‘Am I working on the right things? What direction are my customers’ tastes evolving toward? What interpretation of the data works best for my consumer?’ The next edition of MakerSights will address these questions.”
Specific features include more sophisticated test methodology, including price sensitivity testing and acknowledgments to survey respondents via text message. Improvements to the analytic functions entail a graduated scale of standards evaluating a product as a winner or a loser, and the ability to compare like products on a set of benchmarks. The new version will also be able to analyze a product’s potential to a specific geographic market. Most importantly, it will also have the ability to predict ROI for a given product at any point in the cycle.
Other major brands that have adopted MakerSights include Sephora, Ralph Lauren, MM.LaFleur, Wolverine and Reebok, with more than $4 billion derisked in the past year.
Perhaps MakerSights’ most intangible advantage is the new level of customer loyalty it can engender. Lewand says that while MakerSights has been helpful in making product and marketing decisions, the program has also cemented Shinola’s relationship with its customers.
“We’ve come to see our customers as our partners,” Lewand says. “They feel invested in our brand and are providing us with valuable feedback. That deeper connection is by far the strongest benefit we’ve received from MakerSights.”
MakerSights has also become integral to both the design and strategic sides of Shinola’s business. “There’s a tremendous value to the art and design of the product, so we’re always pushing our team to find that next coolest thing. MakerSights provides useful data to inform the creative process,” he says.
“But when we make related decisions on depth of buy, how we allocate our working capital and how we structure the manufacturing schedule and process, MakerSights’ machine learning algorithms prove enormously helpful. It’s like turning around the rear view mirror to look forward.”
Detroit-based Paul Vachon writes for various trade publications, in addition to feature stories for consumer magazines and books on Michigan history and travel.