MACHINE LEARNING IN RETAIL: AN EDITED APPROACH

Over the last several decades, apparel retailers have struggled with the complexities of increasingly varied product assortments and changes in long established seasonal shopping patterns. As retailers implement new strategies to stay ahead of the demand for an omnichannel shopping experience with emphasis on digital interactions, “they introduce new labor-intensive tasks to sort through all the data they’re collecting”[1].

EDITED, the data analytics company named to Fast Company’s 2014 Most Innovative Companies List [2], tackles this problem through machine learning.

EDITED’s platform crawls brand and retail websites across the globe, monitors consumer opinions on social media, and analyzes output from key industry events, blending machine-learning with human editing. In seconds, it takes vast amounts of real time, measurable data – a task that would be completely unsustainable for a team of human analysts – and turns it into the kind of actionable information that can give brands and retailers a competitive edge when making inventory, pricing, and merchandise management decisions [3].

Since launching in 2009, EDITED’s system has “learned” to recognize apparel products in images and natural language processing software, producing a searchable database of organized, structured information on upwards of 60 million products collected from brands in 30 countries and over 35 languages [4]. Everything from SKU and color options to price and stock levels (as well as many other details) contributes to the formation of one complete data point captured by EDITED’s systems. Processed at a rate of 7 million per day [5], these data points result in a complete profile that allows users to track the evolution of individual items over time.

So what does a brand do with all of this market data and competitor analysis? In the short term, EDITED’s focus is to work with new and existing clients, “start[ing] by analysing their competitors’ historical pricing and assortment data to make more strategic decisions, ultimately leading to better sales, stronger inventory management and less discounting” [6]. Their goal is to build a real time view of the global market leading to more scientific commercial decision-making in the fashion industry.

But the potential for machine learning’s impact in this space extends beyond pricing and promotion. Co-founders Julia Fowler and Geoff Watts acknowledge that their technology has much wider application. In the next decade, EDITED’s machine learning and data analytics will not only enable brands to be first to market with accurately predicted trends, but it will also play a major role in reducing waste industry-wide. The company’s goal is to leverage machine learning to “transform the retail industry by empowering it with the tools to become better, faster, and more efficient” [7].

Long term, EDITED is working to partner with additional brands and retailers to leverage data that will deliver to customers the right products at the right prices at the right time – thereby eliminating over-discounting, promoting customer loyalty, and spurring greater returns and growth.

EDITED has significant potential to increase cost savings, enhance decision-making and encourage process automation. I would also encourage the company to consider how it can go beyond analysis of the current market and leverage its data collection techniques to better understand not only the competitive set, but also the consumer.

In today’s digital economy, customers constantly express their opinions and indicate their preferences online. A tweet. An Instagram like. Clicking add to cart. Each of these presents a host of data points and an opportunity for a brand or retailer to get to know their customer a little better. There has never been more accurate, factual information available with which to measure an industry that is notoriously deemed “fickle.” If EDITED can utilize its platform to “learn” about customer-brand interaction in addition to market analytics, the result will be a more dynamic retail landscape that both predicts and addresses consumer needs.

I would also recommend that EDITED consider the significance of its impact on defining the future merchant role within a retail organization. As EDITED continues to focus on machine learning to produce real-time data, it will not only reduce inventory waste, but will also create greater process efficiencies, the results of which will make for leaner, more streamlined merchant organizations that can do more with less [8]. Over the next several years, the company should work with retailers to leverage the evolution of the merchant role to inform it’s future data collection. Together, EDITED and its partners can best identify what processes might become fully automated and proactively use machine learning to highlight areas where merchants need to focus their time.

But these recommendations demand we reconcile a number of questions surrounding the impact of data-driven intelligence on the fashion industry in the years to come. To what degree should data influence and drive design, buying and merchandising decisions? Should data-driven intelligence ever completely replace human intuition in the fashion retail space? Or is there an optimal mix?

(800 words)

 

Sources

[1] Wilson, J., S. Sachdev, and A. Alter.”How Companies Are Using Machine Learning to Get Faster and More Efficient.” Harvard Business Review Digital Articles.  May 3, 2016. Accessed November 2018.

[2] “The 2014 Top 10 Most Innovative Companies by Sector: Style.” Fast Company. January 01, 2000. Accessed November 2018. https://www.fastcompany.com/most-innovative-companies/2014/sectors/style.

[3] Kansara, Vikram Alexei. “How Realtime Data Is Reshaping the Fashion Business.” The Business of Fashion. August 2011. Accessed November 2018. https://www.businessoffashion.com/articles/long-view/the-long-view-how-realtime-data-is-reshaping-the-fashion-business.

[4] Kansara, Vikram Alexei. “How Realtime Data Is Reshaping the Fashion Business.” The Business of Fashion. August 2011. Accessed November 2018. https://www.businessoffashion.com/articles/long-view/the-long-view-how-realtime-data-is-reshaping-the-fashion-business.

[5] https://edited.com/data/. Accessed November 2018.

[6] Abnett, Kate. “Is Fashion Ready for the AI Revolution?” The Business of Fashion. April 07, 2016. Accessed November 2018. https://www.businessoffashion.com/articles/fashion-tech/is-fashion-ready-for-the-ai-revolution.

[7] “Data Enablers: Apparel Retailers Dress for Success With EDITED Analytics.” PMNTS.com. May 18, 2017. Accessed November 2018.

[8] Begley, Steven, Rich Fox, Gautam Lunawat, and Ian MacKenzie. “How Analytics and Digital Will Drive Next-generation Retail Merchandising.” McKinsey & Company. August 2018. Accessed November 2018. https://www.mckinsey.com/industries/retail/our-insights/how-analytics-and-digital-will-drive-next-generation-retail-merchandising.

 

 

Previous:

Can machine learning take the pain -and cost- out of how we pay for healthcare?

Next:

The London bus is dead. Long live the London bus.

6 thoughts on “MACHINE LEARNING IN RETAIL: AN EDITED APPROACH

  1. This is amazing. I didn’t know about Edited, but I’ll definitely be researching them. I think the concept of bringing it to a more customer-focused approach is huge for me. Our client is everything, I want them deriving most of the value from any technology capex we spend!

  2. The one thing I always struggled with for machine learning for clothing is that clothing is fashion is supposed to be a way to express and present oneself. We spoke about this during the Gap case in marketing but how do you reconcile using fashion to express uniqueness when fashion and apparel companies start to use machine learning and start looking more and more similar to each other? Can you differentiate if you are all using the same trends to predict consumer demand? I completely agree with the point made above of how to work with the future merchant role to solve this problem.

    My other main question would be around the future differentiation of luxury brands who may not use machine learning and continue to use designers to drive style and create a voice. In a world where conforming is less “cool” and having your own voice is more “in,” I wonder if luxury brands will pull more and more ahead in growth and even market share given their further and further differentiation in this area.

  3. A lot of power has been shifted from producers to retailers over the last 15 years with the help of data, proximity to the consumer in general and store brands. Machine learning is the next step of the power grab of retailers.

  4. I loved learning about EDITED’s business model and how it fits within the retail landscape. Echoing some other comments, I’m also curious about how to address the issue of whether creative directors and their “human” intuition will be necessary in the face of technology being able to make intelligent projections of trends. What really resonated with me was the value it brings in waste reduction – H&M was recently accused of burning 12 tons of excess inventory, and Burberry among other luxury brands have also had a history of selling excess stock to preserve premium pricing and allure. Considering the shift to sustainable businesses, I look forward to seeing how companies leverage machine learning technology to better forecast inventory levels.

  5. Thank you so much for sharing this. I had not heard of EDITED before this post and am really inspired by their product! I believe edited can revolutionize the retail industry similar to how social media management programs revolutionized how brands interact with consumers, what data they can collect, etc.

    You bring up a very interesting point about whether these algorithms should be replacing humans. I am a strong believe in automation however I don’t think that EDITED will eliminate an entire data analytics department. I believe that there will be less people needed to run SQL queries, pull data, analyze it, and send off to co-workers. This product will allow less people to be in that position and will open up more roles to help action the data and insights that are being brought forward. In short, I believe this will shift labor resources, not completely replace them!

Leave a comment