The Fashion Retailer

AJ Mak is 34 years old, raised in Hong Kong and Toronto, studied at Carnegie Mellon University in the US.  Worked at one of the largest advertising companies – JWT – before spending 10 years in the apparel industry covering brand management, supply chain management, product development, and sales. Hobbies include basketball, snowboarding, photography and making a difference.

TFR: What is Chain of Demand?

AJ Mak (AJM): Chain of Demand is a predictive analytics company that empowers the retail industry to maximize sales and minimize inventory waste with AI-driven sales predictions.

TFR: What is the entrepreneur story that inspired the launch of the company?

AJM: In the past 3 years, I saw firsthand the effects of the retail apocalypse on brands and retailers. Markdowns and deadstock were destroying margins throughout the industry, which led to products being discarded into landfills and in some instances, burned. It was then that we set out to find a solution to help the retail industry improve sustainability and profitability leveraging the power of AI.

TFR: What are the main challenges in the fashion retail industry in relation to your value proposition or solutions in management, supply chain, buying & planning and product development?

AJM: The main challenge faced by the fashion retail industry is the availability of a solution that is easy to adopt and focuses on the specific pain point in the specific industry.  There are a lot of analytics solutions out there that provide part of a solution to the inventory problem, without offering specific and actionable items. While on the other end of the spectrum, some companies try to solve all of retail’s problems – if you tell them precisely what you want – yet many retailers do not have the time or resources to spec out an entire solution from scratch.

TFR: An issue faced in many projects with Fashion retailers is how disintegrated is planning. It´s very difficult to integrate the value chain, from product development to product distribution when systems and people are working in silos. 

An ideal forecast would analyze the assortment at SKU level in the right store cluster/channel. This could require analyzing data from CRM (Customer Relationship Management system), online traffic, weather or location, plus in-season uncertainty that could generate a sale impact because of a marketing advertising or a celebrity wearing a specific garment. How far are we from integrating those insights to help retailers be more efficient or to design the right product, send it to the right store, place it in the right space and in the right time?

AJM: This has been our goal from the start, and we have built our solution to integrate all those insights to have the right product at the right time, all at the right place.  We essentially funnel a myriad of data sources and distill them into specific action points.

TFR: Another important restriction is that many retailers have low maturity levels of master data management. If a product is not well described, even from the PLM (Product Lifecycle Management system) perspective, it would be difficult to filter the assortment at attribute level (e.g. occasion, fashionability, color, etc). In this sense, many retailers are far from implementing machine learning.

How do you help those companies or what would you suggest to them?

 AJM: We see this with over 90% of companies we work with and to solve this problem we have developed our own AI image recognition system that is able to identify additional features and attributes from product images. This has greatly improved our models’ understanding of individual products and in turn, our prediction ability. 

We do suggest to all companies to start entering as much data into their systems as early as possible. Regardless of which solution a company ends up using, data will be the single largest factor in a successful implementation.

TFR: Could you briefly explain a couple of business cases where Chain of Demand participated? What was the need of your client and how did you solve it?

AJM: Although we are still a young company, we have been privileged to work with Giordano, Bluebell Group (which operates brands such as Jimmy Choo and Moschino in Asia), Petit Bateau, and Hallmark Babies, to name a few.

One specific case study was with London-based ladieswear Jacques Vert. With annual revenues of US$150 million, the brand faced the challenge of efficiently predicting consumer demands quickly. According to the brand director, they only used a merchandising tool and relied on interdepartmental sales meetings to forecast. Like many retailers in the modern-day, they sought for something more powerful and more agile. 

After understanding and aligning with their goals, we pulled data from internal and external data sources – instead of just historical sales – and applied to our many different machine learning models to the data. As a result, our predictions showed we were able to reduce markdowns by 24%, increase full-price sales by 21%, and increase gross margin by 6%. These results were similar to a case study we did with Giordano, where we were able to show a reduction of markdowns by 14%, and increase gross margins by 8%.

TFR: What changed in the retail industry compared to 5-10 years ago and in relation to technology?

AJM: Over the past 5 years, much has changed in retail technology.  There are now a lot more resources in machine learning tools that enable new companies to build industry-specific solutions. Other major changes include the prevalence of cloud storage and substantial increases in internet speeds, which makes for more agile and cost-efficient applications.

TFR: Big data analyst, a specialist in machine learning, developers…is very demanding in the big data era. How do you manage to find these profiles? How do you keep them motivated as Millennials aren´t afraid to change jobs?

AJM: We utilize every possible channel to recruit our talents, and we are doing it constantly.  Keeping our team motivated is an on-going process, and we have found that having a clear company direction, as well as an over-arching aim for creating positive social impact, are the most important factors in attracting and retaining the right talent in our company.

TFR: As a consultant in Accenture, I have been involved in a project about Allocations for an important fashion retailer in South Africa. The challenges the retailer faced went beyond technology or analytics. It´s was about people, processes and systems, and the way they can deal with this complexity. Many executives are slow or reluctant to analytics initiatives. What do you think?

AJM: This very true across many different markets.  In fact, the larger the company, the more difficult it is to drive change throughout its ranks. This was one of the largest concerns we had when planning a B2B strategy, but fortunately, we have been able to get executives to engage in pilots with us in just a few weeks, compared to the average industry of a few months.

We believe this is due to a large number of publications on AI and data analytics, combined with the increasing number of challenges retailers are facing in the past few years. 

One of the largest factors in our short onboarding process is the fact that we had built our system and process to be as simple as possible.  My experience in being on the buy-side of B2B software has helped us streamline our product and presentations for quick buy-ins and proof of concept.

TFR: As a start-up, what do you think are the key success factors that brings you to compete to big players like SAS or analytics departments from large consulting firms?

AJM: One of the key success factors is that we are a very flexible startup. What I mean by this is that we are able to refine our product and solution to fit our client’s needs after each discussion. This is why one of the most important steps in our prediction process is aligning business strategies so that our deliverables solve the agreed pain points.  Another key factor to our success has been how quickly we are able to test, prove and deliver targeted results compared to the industry average.