by Anna Salewski, Founder & CEO at SCROBLE
Part 3 What does data have to do with fashion? Does fashion generally need data if it is mostly about converting the designer’s visions into wearable clothes? In my previous articles, I concluded that the sweeping transformation of consumer expectations and behaviour towards clothes has increased consumers’ expectations of fashion brands. And since modern fashion is pre-produced for future consumption, there should be an urgent need to understand customers’ wishes and desires in advance to satisfy them with appropriate collections to reduce unsold stock.
Innovation and digitalisation providing few tools to help brands understand customers’ needs nowadays. Despite radical digitalisation and innovation being essential for the consumer, brands and retailers who want to implement such advances experience many difficulties. Notwithstanding that we frequently associate innovation in fashion only with e-commerce, AR/VR enabled technologies and textile innovation, in this series of articles, I focus on existing inconvenience and limitations in physical (offline) shopping. I also try to find systematic outcomes of this existing disconnection between online and offline retail and figure out how we may finally improve the status quo as well as hypothesize the exact outcome of such a change.
I would also like to draw the line between data necessary to create and produce collections and data for finding and targeting customers to create (more) sales. In my opinion, the digitalisation of offline retail has great potential to reshape the fashion industry; in this article, I would like to focus more on the role of data for creating collections and understanding customer wishes and expectations in this regard. I aim to closely examine the particular data that could help solve the problems many fashion brands experience with building right perceptions and predictions and enable better planning and later performance. Which specific data would make the most sense in propelling efficiency in all fashion processes? Can we influence sustainability with data? If yes, which data could we use? From where and how do we obtain it?Fashion is and has always been about the plurality of designers’ voices and consumer tastes. Wishes and needs are greatly dependent on geographies, age groups, body shapes and other variables — probably more than, let’s say, the food industry. There are more steps involved in decision-making (observing the fashion trends, finding the best deal and the best fitting garment and others), where tendencies and styles are constantly changing. Because of this plurality of single separated and segmented processes, the role of data for fashion brands should be quintessential, especially for analysis of customer behaviour and the ability to make the right predictions and perceptions. Still, the fundamental implications of big data and the solutions it offers in fashion have been overlooked, precisely due to the complexity of the fashion business and challenges in implementation. Let’s focus more on this point below.
Which data would help fashion brands to comply with customers’ expectations and personalisation and help create garment collections more likely to be sold out?
As I already mentioned in my previous article, I truly believe that every unsold garment is unsustainable per se, no matter how sustainably it was produced or delivered. We can attribute a large part of sustainability problems in fashion to the inability to properly predict what will be consumed and where. In this regard, I think about the relevance of several groups of factors and information which, in my opinion, would be extremely important for fashion brands in order to improve the creation of the collection as well as to gain a strategic understanding of the distributional and personalisation aspects:1. “Product — customer” — relationship data. This concerns why, where and who liked, followed and finally purchased my garment. What is the origin of the interest? Was it spontaneous? Was the customer influenced? If yes, by whom?2. “Product purchase” — decision data. How and why did someone decide to purchase my garments? What was the path and what influenced this decision? Was it price or willingness to purchase a certain article from my brand that tilted the scale?3. “Customer as consumer KYC”. This group of data segments and analyses the consumer of my brand more thoroughly. What are those social, demographic and behavioural patterns leading to my particular brand?
I know it may sound quite far from the existing reality, but I am sure that technological solutions are required to enable this initial and essential product related traction in fashion. Essentially, it concerns the connection between fashion brands and fashion lovers. I am also sure that we need to create a way to receive this information and data in full compliance with personal data protection policies such as GDPR. I also believe that this data needs to be made available to fashion brands.
Which data do we have now? The influence of unproportionally divided consumption in fashion and existing business models.
Due to the complexity of the industry’s distribution channels, and even more due to the absence of suitable technological solutions, it is still almost impossible to capture offline consumer data (which constitutes 80% of total fashion consumption) properly within current forecasting data, thus making it impossible to cater to customers accordingly. Anonymous shopping in city retail is still the norm and widely presented (unless we use loyalty or membership cards). Even when cards are used, what can this information really tell us regarding the points described above? In reality, only small percentage of physical retail purchases are tracked as a simple “events of a sale attributed to an identified client” via client relationship management systems.
Technology has been trying to address the issue outlined above with sophisticated solutions, including AI-equipped cameras able to identify consumers’ facial reactions when looking at a certain garment or categorize consumers depending on their shoes and allocating them a certain style preference and customer personal group to extract valuable data from. Some stores utilising AI assistants with face recognition offer personalised suggestions to known customers based on previous purchases. Other proposed methods to capture data about repeated consumption is by tracking purchasing events through credit card transactions in offline retail. Retailers especially try to attract customers by using location technologies that can show how you and your phone move through the world. Other sources, mostly used by retailers, include social media sites or buying data from ad tech companies.
Even if these solutions may sound interesting, the first critical point is data reliability — facial reactions and shoe choice cannot guarantee a bulletproof result, because these factors can be contradictory or depend on a momentary whim or passing mood, and do not necessarily reflect the style and habits of the observed consumer. They are temporary expressions rather than constantly applicable data attributes. This highly sensitive and very personal data and the ways of collecting it are still widely discussed in the EU. Another point is the data attribution — in my opinion, in my opinion, the extracted information still cannot be interpreted properly, as it is just a small part in the chain of relevant bahaivioral patterns and attributions out of the whole context of interacting with fashion. The existence of these solutions merely, unwillingly, divulges the pressing need for offline consumer data.
Due to irregular patterns of fashion consumption and the absence of clear and unified technological instruments for data collection and capturing offline retail clients, there is a wide limitation in relevant data availability. For fashion brands, there is almost no other way than to apply available online data to the offline world and rely on completed sales data by making new predictions for upcoming collections. Narrowing down the quality of existing data from offline retail available directly to brands, the bitter truth is that more than 90% (my own estimation) of existing data related to customers in physical retail is not useful because of the described quality of information. To explain: First, the traction of offline data is quite limited, as already shown. The small part available contains less relevant information solely related to the specific purchase event or customer behaivior in the store, without focussing on diversified user patterns and nomadic consumer behaviour. But the circulation of behavioural patterns is, in fact, much greater than what is made available by tracking only the single purchase in the store. It is possible to collect and attribute this data online, at least theoretically and when all transactions occur on one e-commerce platform. Another question here would be how can brands access this data? Let’s see more about this later.
But what about consumers leaving stores empty-handed — no one knows why people enter stores and fail to purchase anything. What about that data? In fact, only one out of five customers make a purchase when they visit a store; according to the latest statistics, about 96% of consumers leave stores empty-handed. That means that we (maybe) have a small part of relevant data about one customer who has made a purchase (see above), but nothing about the four others who left the store empty-handed. To conclude, I would estimate that about 99% of relevant data is in fact not available and accessible to anyone in fashion retail.
Suppose we could capture data about everyone who stepped in, left a shop and later made a purchase? What about this gap? It is far more relevant to analyse why a customer was initially interested in a particular product or brand but didn’t end up buying it or the expectations in place while entering a store and what factors were considered during the decision-making process. What if we were able to extract this data in real time and use it to solve the issue through a limited and complicated product search and help customers find what they were looking for?
In fact, and as demonstrated by McKinsey, behavioural patterns are becoming increasingly unpredictable ; people can be influenced by a brand online, but end up purchasing something similar offline or vice versa — they look for something offline and end up making a purchase online because of the variety of offers.Connecting online and offline retail data regarding product–customer behaviour appears to be essential for quality data assessment.
How do brands nowadays obtain data? What is available and how is it used?
There is, as stated above, almost no or no quality offline retail data available directly to brands. The only available and reliable data is still derived from observation of consumer behaviour/purchases/consumed ads online. However, it also bears certain difficulties in interpretation: first, because of online retail diversification and data availability to the brands (because of the plethora of online shops) and second, because of data quality — it is difficult to attribute existing information to customer behavioural patterns or tangible garments.
Fashion brands that produce collections, distribute them and trying to maintain contact with clients rely, of course (depending on a distribution strategy), on many sales channels. There is a lot of literature about challenges for brands to remain profitable and go online, especially through one of the widely used e-omnichannel solutions. By doing so, brands use new e-commerce distribution channels to increase their sales. However, these e-commerce platforms execute their own power by leveraging this customer contact and technology to gain unique data from the source. It is not a secret that one of the biggest values of all e-giants is the amount of data they hold about their customers and product consumption. This also creates a great dilemma regarding gaining and using data in order to build predictions and perceptions, which are important for fashion brands. Applying some mathematics, we reduce again the amount of data available directly to brands, which they would need to make educated decisions for future collections.
The directly available data is later fed into algorithms, which offer the brand some background information. However, again, the type of information available is highly relevant. Data from e-stores of fashion brands capture very limited information regarding customer behaviour, as it is restricted to their own collection within their own store, whereby broader data available to e-multibrand giants is unavailable directly to brands. What a dilemma!
As a logical consequence, fashion brands often use third-party data providers which help them extract relevant data from different sources using AI, image recognition and application of necessary algorithms to develop conclusions regarding upcoming trends and gather observations about competitors. Some fashion brands employ data scientists to obtain relevant data. The giants, such as Nike, even purchase data companies to enforce their understanding of their customers.
However, even having the best data scientists and algorithms will not solve the problem of non-existent offline retail data and its partial inaccessibility (online commerce). There is a tremendous gap in data availability and accessibility which technology can and must fill. Brands should have an immediate better understanding of and direct connection to customers. The ideal would be to empower this connection with different tech features to create a 360° overview of behavioural patterns in fashion.
Is the absence of smart data connected to fashion challenges?
After discussing the various ways in which a customer interacts with fashion, the answer would be, yes, the absence of smart data is partially connected to fashion challenges and plays an active role in maintaining the fashion system’s current position. This absence directly affects production, distribution and logistics decisions, which are connected at the backend to overproduction and unsold stock. None of these factors can be overlooked, as they directly impact profitability and even brand reputation. With the right data at hand, brands can produce what is more likely to be consumed and reduce their dead stock and environmental footprint, thus improving sustainability and profitability. By producing, delivering and selling smarter, brands can automatically increase sustainability.Specific data would allow fashion’s environmental footprint to be further reduced, whilst also allowing for increased profits. Data could even indicate where a brand needs to deliver its products i.e. to the location most favourable for specific products; there would be no need for pointless and costly logistics and reverse-logistics. The role of unique and complete data chains (including consumer patterns and the relation to products) is crucial, especially in these current and upcoming challenging times, when only efficiency will decide who will be successful.
However, my personal conclusion is that existing business models in fashion will ever prevent brands to gain a big part of reliable and relevant product related consumer behaivior data. Notwithstanding the fact that we are swimming in the ocean of the personal data, there is a fundumental difficulty to access and to convert it into valuable analytics for fashion brands.
Is it all only about data?
Nobody wants to promise that data will solve every single fashion challenge. Data collection itself depends on technology and applied business methods and models. However, I wish to open a conversation about the topic. Fashion is a highly consumer behaviour-related industry and must have a way to gain data with respect to consumers’ privacy. Data accessibility will always depend on a business model and tool connecting fashion brands with customers and on whether there is a retailing third party involved. We would need a very smart unified technological solution that can synchronise the entire data circulation to capture different behavioural patterns at different levels of the consumer journey and deliver them directly to a fashion brand. Moreover, this solution must be able to provide the same data and customer connection for offline retail.
Even if every single brand decided to overcome its difficulties and systematically digitalize its inventory and processes, it would not provide the desired customer connection or provide the relevant volume of diversified data to make exact perceptions and predictions about consumer behaviour and upcoming trends. However, whether it’s about creating a collection, rethinking strategy or customer experience solutions — all these activities need to be based on more concrete, relevant and reliable data. This will decide everything, at least in the near future.
So, is it possible to develop something fundamentally disruptive, a product that addresses and solves the multiple needs and demands also in relation to accessing reliable data for fashion?
Photos by Markus Spiske