A research on challenges in fashion buying has revealed that retail buyers face five major challenges in the buying process including trend forecasting, stock maintenance, pricing, vendor selection and market conditions…
New technologies are disrupting the way business is done and the need for adapting these technologies within the business processes and value chain is increasing. The aim of this thesis is to analyse one such business process under the fashion retail industry and explore the use of technology in it.
This study was done in two main parts: literature review and empirical research. The first part of the study analysed the literature on retail industry, fashion retail industry and the retail processes such as buying, merchandising, marketing and range planning. The empirical research investigated the challenges faced by the retail professionals in the fashion retail buying process. Also, the challenges were brainstormed with Artificial Intelligence and Machine Learning experts to suggest a possible solution. The study has been conducted between 2018-2019.
Buying is till date one of the most cumbersome and challenging aspect of the fashion retail industry across the globe, especially at a time when the fashion industry is going through a considerable shift in this era of digitalization. Social media has impacted our society in many ways, yet one of its major contribution has been in fashion and lifestyle. Media, celebrities, influencers, brands and even consumers themselves are contributing to fashion inspirations. This has accelerated the frequency of trends; more communications mean faster changes in fashion and the need to adapt even quicker by the retail brands. This has also had massive impact on consumer behaviour in recent years.
Online data reveals that an average consumer in Southeast Asia spend about eight hours a day on social media, video streaming and online shopping. The availability of such huge content has changed the consumer journey and expectations. The easy availability of information and decision-making power has made the consumers less brand loyal. Among millennials, two-third prefer to switch brands for a discount of 30 percent or more. Their purchasing decisions are also based on company’s practices, mission and values. This shift has forced the companies to be able to deliver convenience, values, newness, quality and price.
With the shift in consumer behaviour it is also essential for fashion retail companies to rethink the way the stock is purchased for their stores. The overall fashion retail buying process also needs to shift from push analogy to pull analogy. This means that rather than buying the stock based on the available products the companies must focus their buying based on the customer demands.
Challenges in Fashion Buying
The findings of the research, based on an online survey and semi-structured interviews conducted in total 21 fashion retail professionals, reveals the challenges faced by them. The research revealed that retail buyers face five major challenges in the buying process including trend forecasting, stock maintenance, pricing, vendor selection and market conditions.
There could be many different individual reasons for these challenges, but as per the study, these reasons could be mostly due to non-precise customer segmentation, under-utilization of existing data and lack of technological advancements in the overall process The most common challenge that was mentioned by these respondents was trend forecasting; analyzing the past sales and future trends to decide what range should be purchased for the store during a season, was challenging for most of the retail buyers.
Another important challenge, overstock is the result of improper customer segmentation and trend forecasting. If the products are not purchased and priced based on customer demands, it is possible that they perform bad during the season leading to overstock situations. To understand how much quantities of what needs to be purchased is an outcome of proper Range Planning, if the ‘What’ has been forecasted properly, the ‘how’ is easier to be predicted. Other challenges such as right pricing, selecting the right vendors and market situation are external factors but can also be dealt with better market and competitor understanding.
For a profitable range it is essential to understand, ‘what to buy’, ‘how much to buy’ and ‘when and from where to buy’. When the buyers are not confident on their customer segmentation, and when they do not forecast properly what to buy for the coming season due to poor trend forecasting methods, it is difficult to estimate the pricing or how much quantity is needed to ensure the right stock. This leads to improper buying and can eventually lead to over or under stock problem. Poor trend forecasting also means that enough market research has not been done on the trends impacting their industry and hence, they are not ready for sudden market changes. Similarly, selecting vendors also depend upon understanding the range, if the buyers know exactly what they want it is easier to select the right vendors.
This shows that the challenges faced by the buyers are somewhat related yet different. But, if they can answer to the ‘What’ correctly, the other answers might be easier to find.
How AI Can Help
The other aspect of the research was to find a possible solution or solutions for the identified challenges using Artificial Intelligence. To fulfil this objective again related theory was studied and in-depth interviews with AI experts were conducted. Based on the previous findings, the most common challenge of trend forecasting was discussed for finding solutions.
A solution of creating an AI Assistant, was proposed to help the retailers and retail buyers. The features of the solution are identified using the literature review and in-depth interview findings and there are lots of possibilities to innovate in this field. Solutions based on Time-Series trend forecasting of future trends and past sales record using Image Recognition systems in Machine learnings was proposed. The predictions based on image recognition from social media and internet would be very beneficial in the fashion industry to understand the trends based on features such as styles, colours, patterns.
In addition, an idea of using the developing technology of Generative Adversarial Networks (GANs) in machine learning can be explored more to help generate images based on the customer preferences helping the buyers to know what customers want. This feature can be incorporated in creating Range Planning boards for the buyers within the AI assistant tool. To enrich the tool, additional features such as image search on online inventory and automatic image tagging for e-commerce sites and inventory uploads can also be included.
The features mentioned above are based on the theory and ideation process. No practical implementation has been done and therefore, how much of this can be achieved within a single tool can only be seen with test and trial. The project can start with one feature first and then eventually adding other aspects to it. Based on the findings, a time-series forecasting of future trends using image recognition of at least one single feature such as colour could be the very first application.
The next phase could be the addition of other features such as styles, patterns, followed by the integration of past sales records and addition of additional features. The step-by -step process of developing this AI Assistant enables quick development of an MVP (Minimum Viable Product) and it can be improved using the feedback received in the early stages.
The possibilities mentioned in this research are backed by the theoretical developments in AI discussed in the literature review. This also fulfil ls the objectives of the research and answers the second part of the research question. New technologies are disrupting the way business is done and the need for adapting these technologies within the business processes and value chain is increasing. The penetration of latest technologies in the fashion industry is not high but is growing and therefore, the need for research in this area could result in innovative solutions, not only in the value chain, but across the industry.