July 11, 2024


Clothes that smile

Digital Transformation in Fashion (4) – part 2/2

Digital Transformation in Fashion (4) – part 2/2

It seems Covid-19 pandemic is ending but uncertainty is the new norm and adapting to it is essential. Fashion, like other industries, is adapting to digital transformation by redefining customer engagement and optimizing operations across new business models with a new approach to product development where sustainability and traceability are mandatory. This article is the continuation of Planning in Fashion, where design meets operations.

Executing the Marketing Mix


An essential step in the merchandise cycle is analyzing product performance. To do that, merchandisers analyse product performance (at different levels across the product hierarchy, usually at style / color level) using attributes and grouping like ABCD, where A are top sellers and D are low selling items (slow-movers). Traditionally, attributes were product-centric, a kind of innate designers/ suppliers language to understand what the features or details of each garment were.

Today, attributes are customer centric (designers are more inspired by short term trends) and technology and especially machine learning enable predictive analytics. An example of product description that is transformed into attributes is “green mint, floral print, short sleeve, shirt” corresponding to the color print and category descriptions. Machine learning deciphers the chaos of data translating it into product attributes. In this case, a “floral green t-shirt”, from the example above, would be a “trendy casual seasonal shirt” or a “beach party shirt” from a customer’s point of view.

AI translates the creative language used by designers into an operational business language that will help planners make better business decisions after analyzing the probabilities of selling this shirt in a specific location, at a specific price and time period. Planners analyze this information to understand shopping missions (eg. work, party, weekend) and optimize the assortment at store level, while eCom platforms display the product ecosystem the visitor is searching for. Retailers manage an unlimited amount of data that features attributes that can be analyzed to support the process of defining “the right product, in the right place, at the right time, at the right price, in the right quantity”.

Demand Forecasting

Demand forecasting is the process of evaluating the number of goods that consumers will probably purchase in the future. Demand is therefore a key indicator for every business and fashion is probably one of the most complex businesses when it comes to demand prediction.

A business that sells commodities can easily forecast demand because demand is stable. In grocery stores, many categories have stable demand patterns even if a volatile event can break the rule (eg. toilet paper spike in sales during covid-19 lockdown). Food retailers have other challenges with regard to planning: optimizing their assortments, dealing with different brands and their private labels, from how deep and wide their selection is, to what the right product placement should be, to the number of product facings and inventory levels.

Food retailers, and DIY ones, use planograms (diagrams that provide details on the placement of every product on a shelf) but their business is very different from fashion. In Fashion, product display and layout are important from a customer experience perspective. But developing space planning capabilities like in grocery stores, has no sense when physical direct goal is not only conversion.

I have worked with many planners in different fashion companies and they usually do their forecast using the last two years data (eg. AW19, AW20) and update their forecast using last season sales, if it’s not a like-for-like period (Latest estimate based on recent sales trend). What surprised me, and I’m talking about companies with >€1B revenues, is the lack of standardization of the forecasting and planning process. Depending on the planner, a different method will be applied and as they usually use excel, it’s difficult to check, see possible errors or have a single version of the truth. This is a manual process where individual decisions are very important.


Why some products are sold in specific store and/or in a certain period of time? When analyzing sales performance, planners look for trends and patterns: like how a product / category (at a color, size, or other factor level) is sold across different store formats, locations or regions. Once the range or line plan is developed, planners (or merchandisers) need to adapt the assortment to a store, store cluster or store chain.

As I previous mentioned, Fashion can take some lessons learned from grocery or supermarket distribution. A planogram is built at category level taking into account many variables such as sales, margins, price perception, product placement, product size, attributes (eg. sustainable, best seller, cros-sell, family-format, etc).

Digitization provides more accurate data and tools to optimize store performance but in order to survive omnichannel requirements, fashion (retail) shouldn’t target the supermarket inventory model. Sales per square foot indicator is not the only metric to measure success in a fashion store. In my opinion, space optimization doesn’t work when apparel retailers are adapting to the omnichannel, redefining the role of the store and improving customer experience. Fashion retail is moving into an entertainment experience: retailtainment.

In Fashion, shopping is not as rational. It’s emotional and aspirational. Customers are looking for experience, socialization, status…however, there are still many things that we can get from grocery. A clear example is the way Grocery companies cluster stores and create different store formats, from supermarkets and hypermarkets to convenience stores. Nowadays, many fashion retailers cluster stores but grocery stores have a much deeper understanding of space planning (eg. store layout, product placement…). Retailers calculate return on space through heat map analytics (cold/hot zones) and KPIs such as conversion rate, average selling ticket, sales per square foot or meter, foot traffic, units per transaction, average units per ticket…but in the digital era where online and offline converge, many metrics become useless or obsolete.

With the emergence of e-commerce, retail brands are transforming their stores into experiential spaces that are part of the customer journey but not the finish line. I mean that the transaction or conversion doesn’t “have to” end in the store anymore. The store is the place where customers can discover products, touch them, see the “real” look, fit to ensure it is the correct fit and size, and then purchase it online. The new omnichannel wording includes BOPIS (Buy Online, Pickup in Store), ROPO (Research Online, Purchase Offline), BORIS (Buy Online, Return In Store)… these new e-commerce acronyms describe how the role of the store has changed.

Leading fashion companies, like Zara or Uniqlo, are already using data analytics and consumer insights. Fashion retailers have higher margins compared to grocery stores and this is one of the reasons they launched e-commerce faster and embraced omnichannel earlier. On the contrary, grocery retailers still have room for improvement in e-commerce and many of them don’t offer home delivery yet. Retail companies tend to segment their stores according to sales, size and inventory turn (stores are graded by A,B,C) even some stores could have special features like a “flagship store”. This segmentation is usually done at product line level (eg. Menswear), but some best-in class retailers make it at department level (eg. Menswear shirts).

On the one hand, clusters are used to manage the quantity of items (at size level) to be sent to each store, but on the other hand, product characteristics are also used to decide what product should be sent to what store (eg. a polo short in a blue color, slim fit with boat detail to be sent to beach shops). In both cases, AI is helping to understand internal data (historical sales, customer profile, sizing…) and external (weather, social media, special events, locations environment…) that will deliver accurate outputs to make better decisions.

Some companies in the fashion industry are already offering curated or store-specific assortments, like Nike by Melrose in Los Angeles. H&M also turns to big data and AI to tailor merchandising mix of individual stores. The company aims to reduce markdowns by using algorithms to analyze store receipts, returns and loyalty-card data. Other retailers adapting their store format to location are Nordstrom, IKEA or Decathlon.


In retail, pricing is the single most impactful lever for profitability, but was probably one of the most underutilized ones in the fashion business. Pricing has evolved as well at the pace of new improvements in technology. Traditional pricing was based on cost of production (cost+margin) and competition. Companies used different pricing strategies such as cost-based, value-based, everyday low prices, skimming or penetration pricing, amongst others.

In fashion, planners and buyers agree on a pre-season price based on cost and competition, to mention a few factors. As mentioned in the introduction, retail companies’ operations and their network (brands, suppliers, distribution) were based on a linear and sequential supply chain model. The retail calendar was very basic, meaning that there was a full-price sales season and a sales period. In some countries, government regulations restricted promotions or special marketing campaigns. As a result, the product life cycle curve that goes from introduction, growth, maturity and decline phases had a predictable pattern. Pricing was simple: making changes at the pace of product life cycles and seasonality and aligned to the marketing / retail calendar, that in many countries was regulated (eg. specific rules on price reduction announcement).

In fashion, in-season price management is probably at an initial maturity level with a few exceptions. Pricing, as well many other processes, saw complexity increase when companies expanded to other markets. H&M, for example, is present in 74 markets. As they mention in their annual report, there are risks and uncertainties related to the shift in the industry, fashion, weather conditions, macroeconomics and geopolitical events, sustainability issues, foreign currencies, taxes and various regulations, but also in connection with expansion into new markets, the launch of new concepts and how the brand is managed.

These factors are impacting demand, inventory levels, costs and margins in real-time. Market deregulations, or new regulations, e-commerce and evolving shopping trends are increasing risks. As a consequence, in-season pricing capabilities are crucial to stay competitive. Planning becomes a just-in-time decision-making process. From a business operations point of view, it means that planners need to deal with multiple changing factors, as well as understanding every specific markets requirements like marketing calendars (eg. Christmas, Diwali, Ramadan, Chinese New Year…). Hypermarkets and grocery stores were probably the first ones to build advanced pricing capabilities in retail. In such a low-margin and high-inventory turnover industry, grocery players are fighting for market share. They implemented dynamic pricing solutions aggregating customer trends, competitors pricing, inventory data and price elasticity analysis in real time.

When shopping in a supermarket a customer decision tree is influenced by factors such as brand, format, size or price. Depending on the customer, the priorities within this tree will change. As an example, when selling wine to a potential consumer, retailers need to understand what the factors are affecting shopping decisions like the type of wine (red, white, rosé…), budget, country of origin, type of grape, wine label design, etc. A customer will compare many different factors, but price is probably the more sensitive one when shoppers are standing in front of the shelves.

Another important point with regard to pricing and customer segmentation is market atomization. While in the past, most companies grouped their customers by segments based on common characteristics, today market atomization is focusing down to the level of the individual. The latest innovation in grocery is sending hyper-customized proximity-based promotions. Retailers know exactly where the customer is in the store and will send real-time promotions to increase cross-selling, basket size, value, etc. Customers compare different choices when standing in front of supermarket shelves. But consumers shopping for apparel in a physical store don’t compare prices in the same way. Generally, grocery shoppers will end up buying the products, while apparel shoppers might well postpone the decision.

Digitalization is changing shopping behaviors. Consumers compare prices using real-time data and check what the right size is in a physical store (showrooming). Customers also know that a branded store or site usually has higher prices than other distribution channels so they will visit the brand’s site to get the right information and experience, but will find the cheapest price on the internet. This is one of the reasons why brands are developing loyalty programs, hypersegmented promotional campaigns and implementing the latest Customer Relationship Management (CRM) solutions. Brands are going B2C and customer information is a key business asset (eg. Nike’s Consumer Direct Offense).

Big data and advanced analytics are giving retailers the chance to adapt to this new way of shopping and price is more important in fashion than ever nowadays that there is such uncertainty with raw materials and logistics costs, changes in FX, etc. Merchandising is a key pillar within the fashion retail value chain that is becoming increasingly data-driven to improve decision making.

*This article is based on Fashion Goes Tech, were processes and examples are more detailed.