Burberry: Digitizing Luxury Retail with Machine Learning
Advances in machine learning (ML) are tilting the playing field in consumer retail. On the one hand, tech-savvy pioneers like Amazon and Glossier are leveraging ML to captivate Millennials and Gen Z consumers with ever more engaging and personalized online shopping experiences. On the other, traditional retail brands, especially luxury labels, hesitate to embrace ecommerce for fear of losing their exclusive, white-glove service experiences delivered through the brick-and-mortar channel. Using Burberry as a case study, this article discusses why luxury retail brands must develop a robust ecommerce strategy to stay relevant, and how leveraging ML can help them win over online shoppers without diluting brand equity.
Positioning Luxury Retail for a Digital Future
The world is becoming increasingly digital. Luxury retail is no exception. By 2025, nearly 20% of luxury sales will occur online, and 80% of the purchase decisions will be influenced digitally1(Exhibit 1). Tech-savvy Millennial and Gen Z consumers will account for 45% of the global luxury market then2. Research on purchasing behavior of this group suggests that their key purchase criteria include: personalization, resonance with brand value, and experience over possession3. These trends hold two important implications for the future of luxury retail. First, having a robust ecommerce presence will not be optional, but a must-have. Second, to win, brands must not only know a great deal about their consumers, but they must also translate that consumer knowledge into products and services that are personalized, responsive and relatable. This is where ML comes into play.
Machine Learning in Retail
In its simplest form, ML algorithms identify patterns in large datasets and use them to generate predictions. In the world of retail, the most common applications are personalized product recommendations and marketing campaigns. Amazon generates 35% of total sales from personalized recommendations4, which are created based on browsing and purchase histories of both the individual shoppers and those with similar shopping patterns. Similarly, digitally-native beauty brand Glossier mines fan comments on its beauty blog to inform new product development and launch decisions5. Tumi, a high-end luggage brand, uses ML to customize its outbound marketing campaigns (e.g., emails and 1-on-1 chats) based on a connected database of emails, social media activities and browsing across the web6. Across these businesses, ML creates a competitive advantage in how they acquire, retain and increase lifetime value of consumers.
Burberry: A Case Study
Whereas most luxury brands hesitate to fully embrace digital, Burberry has made deliberate decisions to invest in and integrate ML into its digital strategy.
Since 2006, the British fashion label has been offering data-driven personalized product recommendations, both online and in-store7. These programs had allegedly led to 50% increase in repeat purchases by 2015. Burberry launched Facebook chatbots during the 2016 London Fashion Week. Like Amazon’s Alexa, these “smart assistants” offered dynamic 1-on-1 interactions with patrons, with key functionalities including selling products from the latest collection and showing behind-the-scene inspirations8. Though rudimentary, the chatbot exhibited abilities to respond to user-generated phrases beyond pre-set buttons, indicating integration of natural language processing capabilities (See Screenshot in Exhibit 2).
Looking ahead, continuous improvements to the ML algorithms require large amounts of high quality training data. To elicit voluntary data sharing from its online community, Burberry has developed an advanced data platform integrated with Facebook and Twitter, to which consumers are encouraged to upload photos of themselves in Burberry products9. These data will enable the brand to further customize the products and experiences they offer. For the longer term, Burberry has announced plans to continue investing in ML across front- and back-end functions. The company’s SVP of IT discussed plans to use ML to automate supporting functions (e.g., development and operations, testing), improve scenario modeling for planning and logistics, and improve security and fraud prevention through ML applications10.
Beyond these planned initiatives, I would argue there is space for Burberry to think outside the box even more with regards to potential ML applications. Some considerations below:
- Ideation: Similar to Glossier, use user-generated data to guide and inform new product development pipeline.
- Immersive ecommerce: Combine chatbot technology with VR/AR applications to create a 100% personalized, immersive digital store, where virtual shopping assistants can replicate the in-store service experience for consumers at home.
- Full product personalization: Based on personal dimensional data and expressed historical preferences, create demos of unique, individually-designed products. Offer exclusively to high-spenders. Produce on a made-to-order basis.
- Supply chain management: Forecast demand at the SKU level to limit inventory pressure and better manage vendor relationships. Wayfair has demonstrated success in this regard11.
- Pricing: Pricing is another obvious area of ML application. However, since luxury brands generally adopt a no-discount strategy, I would deprioritize this lever.
In light of the trends and ideas suggested so far, the following questions merit more thought:
- How should luxury retail brands decide whether to develop ML capabilities in-house or outsource to a 3rd party? If outsourced, how to mitigate concerns about data privacy (both to protect consumers and to preserve brands’ competitive advantage)?
- How to accelerate the process of data collection for training and improving algorithms without compromising data quality (e.g., personal, interconnected, comprehensive, accurate)?
Word count: 741
References:
1. Achille, A., Marchessou, S. and Remy, N. (2018). Luxury in the age of digital Darwinism. [online] McKinsey & Company. Available at: https://www.mckinsey.com/industries/retail/our-insights/luxury-in-the-age-of-digital-darwinism [Accessed 10 Nov. 2018].
2. D’Arpizio, C. (2018). Spring Luxury Update. [online] Bain. Available at: https://www.bain.com/about/media-center/press-releases/2017/global-personal-luxury-goods-market-expected-to-grow-by-2-4-percent/ [Accessed 10 Nov. 2018].
10. AI Business. (2018). Where are Burberry with AI? Exclusive Interview with David Harris, SVP of IT. [online] Available at: https://aibusiness.com/where-are-burberry-with-ai-exclusive-interview-with-david-harris-svp-of-it/ [Accessed 10 Nov. 2018].
3. Woo, A. (2018). Understanding The Research On Millennial Shopping Behaviors. [online] Forbes. Available at: https://www.forbes.com/sites/forbesagencycouncil/2018/06/04/understanding-the-research-on-millennial-shopping-behaviors/ [Accessed 10 Nov. 2018].
4. MacKenzie, I., Meyer, C. and Noble, S. (2018). How retailers can keep up with consumers. [online] McKinsey & Company. Available at: https://www.mckinsey.com/industries/retail/our-insights/how-retailers-can-keep-up-with-consumers [Accessed 10 Nov. 2018].
5. Milnes, H. (2018). How Glossier uses data to make content and commerce work. [online] Digiday. Available at: https://digiday.com/marketing/glossier-uses-data-make-content-commerce-work/ [Accessed 10 Nov. 2018].
6. Milnes, H. (2018). How Tumi is using AI in marketing campaigns, online and in stores. [online] Digiday. Available at: https://digiday.com/marketing/tumi-using-ai-marketing-campaigns-online-stores/ [Accessed 10 Nov. 2018].
7. Marr, B. (2018). The Amazing Ways Burberry Is Using Artificial Intelligence And Big Data To Drive Success. [online] Forbes. Available at: https://www.forbes.com/sites/bernardmarr/2017/09/25/the-amazing-ways-burberry-is-using-artificial-intelligence-and-big-data-to-drive-success/ [Accessed 10 Nov. 2018].
8. Maruti Techlabs. (2018). Chatbots as your Fashion Adviser. [online] Available at: https://www.marutitech.com/chatbots-as-your-fashion-adviser/ [Accessed 10 Nov. 2018].
9. Mittal, S. (2018). How To Leverage Digital Tech To Drive Revenue Growth. [online] Forbes. Available at: https://www.forbes.com/sites/forbescommunicationscouncil/2018/10/02/how-to-leverage-digital-tech-to-drive-revenue-growth/ [Accessed 10 Nov. 2018].
11. Supply Chain 247. (2018). Machine Learning Steps Up Retail Performance. [online] Available at: https://www.supplychain247.com/paper/machine_learning_steps_up_retail_performance [Accessed 10 Nov. 2018].
I believe that Burberry should develop ML capabilities in-house. The retail landscape is drastically changing and digitization and personalization are necessary to compete in the market. Since this skill set is something that will be imperative to Burberry’s success in the future, I think they need to invest in bringing talent in-house that can help give them a competitive advantage. This reminds me a lot of Walmart’s decision to buy digitally native companies such as Bonobos and Jet.com. Walmart realized they needed to embrace e-commerce in order to succeed and so they acquired digitally native e-commerce companies—they then can learn from these companies about their e-commerce strategies and apply them to Walmart. Outsourcing the ML job may help in the short term, but for long term success I would advocate for in-house ML at Burberry.
Great article Charlotte – and the question you posed regarding whether to develop ML in house is really interesting. I would argue that Burberry should invest now in developing capabilities to develop ML in-house as opposed to outsourcing. Two reasons – first, the Company would be able to react real time to data as it flows in, and developing the ML muscle will allow the company to utilize the algorithm and data collected more effectively. Second, more than ever, competitors are trying to gain an edge on collecting the right data to improve their customer segmentation and increase revenues; the possibility of outsourced ML data falling into the wrong hands is not worth the risk.
Regarding your second question, I do think quality is incredibly important, especially to a luxury retail brand. Accuracy of the data is also key, as these luxury retail houses don’t subscribe to constant change in styles and collections that are created in a Fall or Spring collection have much slower turnovers. Burberry’s margin of error that it can afford is much smaller than those of fast fashion houses as well. I would encourage Burberry to focus on increasing accuracy and quality over speed instead of finding ways to accelerate ML data collection.
Charlotte – I really enjoyed reading this article. It is clear that is has been very beneficial for Burberry to integrate ML into its digital strategy, and you laid out a clear and convincing argument. In response to your second question regarding accelerating the process of data collection vs. maintaining high data quality, I would argue that the Company should prioritize quality at the cost of speed. Given that Burberry is a luxury brand, any perceived deterioration in quality could have a significant negative consequence on its brand image. For this reason, I would encourage the Company to continue to expand its ML applications, but in a slow and controlled manner.
One additional question I had for you is in response to your proposal for the Company to include immersive e-commerce. Do you think creating a digital store where virtual shopping assistants replace the in-store service could have a negative impact on its perceived luxury brand? Do you think the virtual assistants would truly be able to replicate the in-store service currently offered?
Thanks for the interesting article Charlotte! Regarding your question, I agree with some of the comments above that Burberry should continue to develop its machine learning capabilities in-house. Given the data and knowledge they have of their customers, they are in the best position to tailor to their specific needs, and in the competitive industry they are at, they need unique sources of comparative advantage.
One of my concerns if Burberry continues to move to machine-learning for customer interaction is whether it will start losing its luxury appeal? In a more and more digitilized world, personal interactions can become more valuable. People who buy luxury brands are also buying into the experience, and receiving impersonal messages might deter from this.
This is a great piece on a retailer successfully integrating machine learning techniques into their business. Burberry is in the special spot of being digitally forward, while many others do not have this choice. This article talks about how digital is built into their culture and is not just projects they pursue for short term change: https://digiday.com/marketing/burberry-became-top-digital-luxury-brand/.
Your question about the security of third parties holding data also merits a lot more thought. Retailers have traditionally faced the challenge of having many digital initiatives they want to take on while not being able to hire enough technical talent in house. Thus, in order for most retailers to succeed at truly personalizing the purchase experience, they need to really invest in understanding security requirements and bringing on the right external expertise to help them accomplish their goals. Thanks for sharing!
Thanks for the great read Charlotte! Your first question is particularly thought provoking as every company that is considering utilizing machine learning must be wrestling with the tradeoffs between developing internally and outsourcing to a 3rd party. Personally, I would outsource the development of the machine learning algorithm to a 3rd party for a few reasons: (1) Burberry’s core business is product design and it would be difficult to structurally change the organization to become a technology company, (2) 3rd party vendors have likely completed multiple machine learning algorithm implementations and can leverage prior experience, and (3) the war for talent (particularly for data scientists and engineers) is fierce and I struggle to believe that Burberry would win this battle.
One additional question that came to mind for me was – how can machine learning be used to drive traffic to physical stores? With approximately 240 retail locations[1], Burberry is still highly levered to physical retail and I wonder what applications could be developed to make the in-store experience more interactive and personalized. Do you know of any competitors focused on machine learning applications in stores (as opposed to online)?
[1] https://www.burberryplc.com/en/investors/annual-report.html