Background and Introduction
Zara is one of the leading companies in clothing retail. The company provides clothing for men, women, and children, as well as shoes and accessories. Zara was founded as a family business in 1975 in Spain by Amancio Ortega and RosalĂa Mera. In 1985, Zara incorporated Spanish multinational clothing company Inditex, which allowed the brand to access a distribution system capable of adapting quickly to changes in the market (Roll, 2021, para. 4). Today, there are 2,047 Zara brand stores around the world and 507 Zara Home stores (Inditex around the world, 2021). To provide customers with a wide range of clothing styles, Zara must closely monitor fashion trends and demand changes.
The success of Zara has been attributed to the company being one of the first in the world to adopt the fast-fashion strategy. At the time of Zaraâs founding, it would take on average, six months for a company to bring a new design out on the market. This delay created a notable gap in the market, which Zara aimed to fill: the demand for the latest fashion trends at affordable prices. Ortega developed a new design and production strategy, which focused not on individual designers but on groups. The new approach was aimed at keeping up with fashion trends with little to no delay and was dubbed âinstant fashionâ by Ortega (Roll, 2021, para.4). Several multinational companies now employ this business strategy, dominating clothing retail. This report addresses the different types of demand the company must manage and the strategies to forecast the demand and manage supply. Both the demand for clothing and for home goods supplied by Zara Home are discussed. Forecast strategies are compared and proposed for both seasonal and steady demand.
Main Discussion
Types of Demand
Zara manages the demand for two categories of products: clothing and accessories and home goods (Zara Home). The demand for clothing and accessories has two main components: steady demand for basic goods, which remains largely consistent, with slow changes year to year affecting consumer style preferences. Still, the demand for such products is affected by seasons in most countries. The other component includes fashion products, the demand for which changes rapidly, dependent on several types of trends. These include global trends â seasonal changes and overall fashion trends, and local trends â celebrations and events typical for a particular country. Generally, products go through three stages in their life cycle â growth, maturity, and decline. The maturity of fashion items is long, so the demand for these products is usually ramp-type (Dai, Aqlan, and Gao, 2017, p.142). This pattern corresponds to the gradual increase in customer understanding accompanied by the growth in demand. As the customersâ understanding becomes stable, the demand also flattens (Dai, Aqlan, and Gao, 2017, p.142). However, these long-time trends are not the only demand patterns that Zara has to manage.
The business model in Zara is based on keeping the inventory low at all times. Among other effects, this approach creates a manufactured scarcity, prompting the customers to buy products that may not be available later (Roll, 2021, para.20). The inventory in each store is updated bi-weekly so that the customers are motivated to visit often and find something new every time they visit (Roll, 2021, para.31). This inventory update model puts additional pressure on demand managing and forecasting. To match the demand, Zara manages its supply simultaneously at three levels: the global level, âfamilyâ level, and individual shop level (Aftab, 2018). On a global level, Zara focuses on long-term planning based on sales history, primarily during the prior year. Traditionally, two major collection updates are displayed around Easter time for the Summer Collection and in September for the Winter Collection. At the âfamilyâ level, which groups together shops within one country or a region, these timings are adjusted to take advantage of local holidays and events. To manage specifically inventory and demand in individual stores, Zara employs two types of forecasting. The first type is the prediction of the overall demand in the store for a 4- to 6-month time period, based on the history of sales. The second type is the prediction of demand for individual products, used to plan the bi-weekly restock (Aftab, 2018). Thus, the short-term local demand Zara must address is the second type of demand.
Finally, a separate type of demand is the demand for home goods, fulfilled by Zara Home. This demand is independent of fashion trends but is affected by interior decoration trends. Still, much fewer people follow these trends, making the demand more stable than clothing and accessories (Utz, 2021). Similar to clothing, the demand for household items shifts a few times a year in response to seasonality and local events and festivals. Also, as in clothing, different types of products have different demand patterns. On the one hand, basic products, such as bedsheets, curtains, and neutral décor, have a steady demand, only somewhat affected by global trends in interior design. On the other hand, the demand for such products as seasonal decorations (e.g., Christmas or Halloween decor) depends largely on the time of year.
Managing Predictable Variability
As a fashion company, Zara must accurately predict the demand on all levels to remain current with trends and competitive on the market. State-of-the-art forecasting methods used in the fashion industry rely on the Machine Learning (ML) approach to predict the patterns in demand. Other forecasting methods implement Big Data, Data Mining, Greedy methods, or a combination of them. Each of these approaches has advantages and disadvantages and can be applied effectively to a particular forecasting aspect. Big Data was shown to be particularly effective in analyzing Social Media data and customer feedback (Chang et al., 2020, p.39). Volumes of such data are enormous and can only be analyzed with the Big Data approach to provide insight into customer preferences and their impact on demand and industry. Data Mining can be used to analyze customer data and identify the product attributes most attractive to the customers (Chang et al., 2020, p.39). Overall, ML is by far the most popular approach for forecasting.
For forecasting in the fashion industry, several ML methods can be proposed. First, Sketch-Product Fashion Retrieval Model based on the Deep Learning approach and Convolutional Neural Network can be implemented to develop a recommendation system (Chang et al., 2020, p.41). Singh et al. (2019) proposed a forecasting model based on Deep Learning and a Tree-based algorithm, which takes into account seasonality and trends. Notably, this model also includes cannibalization of the products, which is a significant part of Zaraâs demand management (Aftab, 2018, p. 215). Considering Zaraâs fast-changing inventory, the approach proposed by Tarallo et al. (2019) may be beneficial. This approach is based on Autoregressive Integrated Moving Average Model and is specifically tailored to forecasting in industries with fast-moving products. Finally, several methods based on the Neural Network approach have been developed for forecasting, specifically to construct an apparel recommendation system (Chang et al., 2020, p.41). Specifically, since Zara manages a vast range of products, an ML-based system for product categorization, such as the one proposed by Donati et al. (2019), would be instrumental in product-specific forecasting.
Proposed Forecasting Methods
Considering the scale and complexity of Zaraâs operation, a combination of methods should be applied for effective forecasting. First, the categorization of products and customers should be performed and regularly updated to compartmentalize the forecasting. Second, Big Data and Deep Learning approaches should be implemented to analyze sales history and Social Media data and monitor global trends. Next, a combination of ML approaches should be implemented to predict the demand separately at different levels of operation. Global and seasonal trends are to be predicted for the shops within one country or region; local trends and individual shopsâ sales history must be monitored to guide the bi-weekly restocks. Finally, in the context of the global pandemic, similar ML approaches can be implemented to track the changing regulations in different regions, thus predicting the ratio between online sales and offline retail and managing inventory and personnel accordingly. Such force-majeure events are largely unpredictable, but their effect on demand can be forecasted within certain limits. With the end of the pandemic, the ratio between online and offline retail will likely become more stable, and the company would need to adjust the forecasting methods accordingly. Otherwise, the future changes in the forecasting approaches will largely be shaped by the availability of new techniques, most likely stemming from the rapidly developing ML approach.
Conclusion
Zara is a fashion company that has to respond very quickly to the changes in demand, both due to the nature of the industry and the fast-fashion business model. The company produces a wide range of goods, some with relatively stable supply and some significantly affected by short-term trends. Modern ML approaches are implemented for forecasting in most industries. Considering the scope of Zara, a combination of ML approaches is required for effective demand management at all levels.
Reference List
Aftab, M.A., Yuanjian, Q., Kabir, N. and Barua, Z. (2018) âSuper responsive supply chain: the case of Spanish fast fashion retailer Inditex-Zaraâ, International Journal of Business and Management, 13(5), pp. 212-227.
Chang, A. A., Ramadhan, J. F., Adnan, Z. K. S., Kanigoro, B., and Irwansyah, E. (2021) âFashion trend forecasting using Machine Learning techniques: a reviewâ, in Silhavy R., Silhavy P., and Prokopova Z. (eds) Data science and intelligent systems. Springer, Cham, pp. 34-44.
Dai, Z., Aqlan, F., and Gao, K. (2017) âOptimizing multi-echelon inventory with three types of demand in the supply chainâ, Transportation Research Part E: Logistics and Transportation Review, 107, pp. 141-177.
Donati, L., Iotti, E., Mordonini, G., and Prati, A. (2019) âFashion product classification through deep learning and computer visionâ, Applied Sciences, 9(7), p.1385.
Inditex around the world. (2021) Web.
Roll, M. (2021) âThe secret of Zaraâs success: a culture of customer co-creationâ, Business, Brands & Leadership, Web.
Singh, P. K., Gupta, Y., Jha, N., and Rajan, A. (2019) âFashion retail: forecasting demand for new itemsâ, arXiv preprint, arXiv:1907.01960.
Tarallo, E., Akabane, G. K., Shimabukuro, C. I., Mello, J., and Amancio, D. (2019) âMachine Learning in predicting demand for fast-moving consumer goods: an exploratory researchâ, IFAC-PapersOnLine, 52(13), pp. 737-742.
Utz, C. (2021) âIs it finally time to forget about trends in interior design?â, My Domaine. Web.