How TechStyle Uses Machine Learning for Personalization: Q&A; With Danielle Boeglin

“TechStyle brands utilize machine learning to cluster similar products based on different attributes, such as color, fabric, heel height, review information, etc., which then match customer clusters with product clusters to show users the most relevant information.”

This Los-Angeles based, avant-garde fashion group, which includes brands like, ShoeDazzle, Fabletics and JustFab, seamlessly ties personalization and vertical integration to create a winning combination of brick-and-mortar stores with online retail. We spoke to Danielle Boeglin, vice president of data analytics at TechStyle Fashion Group, to understand how the company marries data and fashion so successfully.

From her approach toward customer experience to the role of AI and ML in personalization, Boeglin shares the most useful lessons for marketers in the digital retail space. In this exclusive with Martech Advisor, she throws light on TechStyle’s experience with integrating the various components of their martech stack, and more.

Key takeaways from this Q&A; on machine learning for personalization:

• Insights on the strategies that work best to target your audiences
• Top tips for building customer experiences in real-time
• Latest trends to follow in machine learning for CX for 2020 and beyond
   

Highlights of the Q&A; With Danielle Boeglin

Here’s the edited transcript of Boeglin’s conversation with us on the importance of machine learning for CX:

TechStyle uses technology, personalization and vertical integration to create a winning combination of brick-and-mortar stores with online retail. What have been the biggest milestones in this journey? What are the building blocks to successfully marry tech with fashion?

TechStyle was founded on the idea that efficiencies in retail can be realized if brands better understand their customers. To do this, we pioneered the use of introductory customer quizzes and almost a decade later the information gleaned from these questions continues to lay the foundation of our business.

• The quiz helps TechStyle personalize the customer experience from the moment shopping with one of our brands begins by considering the preferences and product desires they’ve opted to share with us.
• The development of FashionOS, our proprietary software suite we build in-house to power our brands, was another game changer for the business.

• Expanding into brick-and-mortar retail was another major building block. As part of our entry into retail, we developed and launched OmniSuite under FashionOS to seamlessly integrate the online and retail shopping experiences.

Learn More: How to Galvanize Audiences With Interactive Content Marketing: Q&A; With TCS’ Sunil Karkera

Could you share, with examples, your approach to customer acquisition? What strategies work best in your target audiences, and which tools/ technology platforms are necessary to execute these plans? What are your most useful lessons around customer acquisition in the digital space?

Our approach to customer acquisition is to surface visually appealing creative to the right audience at the right time. In our creatives, we have found that there is an essential mix to captivate our customers and drive them to sales. It begins with highlighting the product features that differentiate it from competitors, such as cellphone pockets in leggings and studded straps in shoes. It then involves featuring the value and quality of the product, finally giving our customers incentive to purchase with an offer or promotion in brand advertisements.

Our brands also lean heavily into the machine learning algorithms of partners, such as Google and Facebook, and their many signals to find the right audiences. Since almost all digital ecommerce companies use these tech partners in their advertising tactics, Facebook and Google have collected over 200+ data points on their users, which then inform which users are most likely to purchase from a brand.

TechStyle brands also use broad targeting based off prior purchase information to inform similar products that customers may want to buy. This way, two consumers of the same age living in the same area will be served completely different ads customized to their personal style. For smaller partners whose machine learning algorithms are not as strong, brands depend on interest and behavioral targeting. Our teams have conducted many market-research studies to learn about shopping behaviors and interests to discern how that may affect purchase decisions.  

Over the past two years we’ve significantly increased our work with influencers and have seen huge ROI. Influencers unlock new audiences that digital advertising can’t always reach. The power of influencer marketing as it relates to the new generation of consumers is unmatched. Young consumers are extremely loyal to the influencers they follow, as it creates a sense of trust between a consumer and an individual that they revere.

But most importantly, creative still matters. Targeting is important but targeting will not drive sales if it does not resonate with the consumer. Brands should be constantly testing creative, noting where they are falling short, and fighting hard to avoid creative burnout. Additionally, brands should make sure they are fully diversified across all ad formats, while constantly updating and refreshing creative to keep it from becoming just another tired ad.  

Which key data points does TechStyle use to personalize its customer experience – be it the website experience or the purchase experience? Does AI and ML play a part in such personalization?  How do you feed performance measurement and analytics back into building customer experiences in real-time?

TechStyle brands utilize the introductory quiz as a personalization tool to determine a personal style for each customer. The quiz informs a behavioral profile, which then informs the products highlighted for that member or customer. For example, on a typical Fabletics quiz, customers are asked where they like to work out, what their favorite type of workout is, their body type and their favorite color palettes, all to inform what products will be displayed to them and recommended. Therefore, a customer who prefers boxing in a gym but loves blue will receive completely different recommendations than a customer who loves doing yoga at home but also loves blue.

Machine learning has a huge influencer in these predictions, as they cluster customers who are like one another based on their shopping behavior, previous click and purchase information, quiz answers and any other attributes. TechStyle brands utilize machine learning to cluster similar products based on different attributes, such as color, fabric, heel height, review information, etc., which then match customer clusters with product clusters to show users the most relevant information.

Integrating the martech tech stack is the number one challenge for most marketers. What has been your experience with integrating the various components of your martech stack? What tips do you have for marketers investing in new martech, when it comes to factoring in integration from the start?

When considering new tech partners, their ability to plug into and seamlessly integrate with our existing tools and platform is our top priority. We pull data from our partners into our own datasets and dashboards to more effectively optimize media campaigns and strategies as well as to feed our internal attribution model. TechStyle chose to build many of these integrations in-house at the time because our teams were not satisfied with the other third-party solutions in the market.

However, in recent years we’ve seen several interesting martech companies emerge that offer the integration we need. When thinking about how to best integrate your martech stack, brands should think about the types of resources that they can put towards support. If brands have small budgets for developers, then utilizing a third-party solution is likely the best path for that brand. Conversely, if a brand has the budget to invest in their own internal resources, this will give them a better opportunity to exhibit more control and flexibility with integrations.

Brands want to think about their martech stack and platform across the following categories:

• Data Management Platforms
• Analytics Reporting & Visualization Tool
• Customer Relationship Management and Automation
• Digital Asset Management
• Social Media/Influencer Management
   

Finally, when considering all these tools, brands should determine how to form integrations between them. They should seek partners that offer those integrations from the outset.

Learn More: Why Online Communities Are a Key Driver of Customer Experience (CX): Q&A; With Higher Logic’s Kevin Boyce

What are your top 3 tips for digital native brands who are just starting to explore influencer marketing?

When strategizing the right influencer marketing campaign, I would offer the following three tips: define your target audience and goals of the campaign; develop guidelines while allowing them creative freedom; and build longer term relationships.

When defining a target audience, brands want to think about the target age, interests, etc. By taking that initial step of defining a target audience, brands will be able to easily identify the influencers that will make for a successful partnership. Does their audience and network of followers equal the audience you as a brand are trying to reach? That’s a very important factor that should be considered early on. Next, brands should determine the goal of their campaign. Depending on the campaign goal, brands will provide influencers with specific talking points, different posting specifications, and will ultimately utilize different measurement strategies.  

Giving influencers the flexibility to be their creative selves is so important. They are content creators and have the followers they do for a reason. Brands should respect that and allow them the freedom they need to maintain the engagement of their following.

Lastly, we have found that when our brands form longer term relationships with these influencers, specifically over a period of six months or longer, the campaign is ultimately more effective than paying an influencer for a single post or for a short-term engagement. We feel that this is because their audience sees the TechStyle brand integrated into their everyday life, so the trust factor is vastly improved.

What do you think legacy brands can learn from digital native and D2C brands when it comes to being in touch with customers’ needs, and engaging them through their journey?

Digital brands focus a lot of their attention on the end-to-end customer experience, making sure to eliminate any friction points, streamlining the checkout process and surfacing relevant product information wherever possible.

Legacy brands should begin to use data as a method to making better business decisions and creating a personalized shopping experience across channels. With digital ecommerce brands, there is access to much richer and granular data on customers than is available to traditional brick and mortar retailers. Collecting data through machine learning is how Fabletics was able to successfully transition from ecommerce to brick-and-mortar, ultimately creating a positive in-store experience that complements the digital shopping journey.

Media reports have often overblown the “retail apocalypse,” leading consumers to believe all physical stores must be dying. But what is dying are brands that don’t understand their consumers. Digitally native brands that expand into physical retail understand how to harness digital data and translate that into an in-store experience, which is what legacy brands have traditionally lacked.

Another piece of advice I would offer is to upsell whenever possible. Brick-and-mortar retailers can dedicate time to train their sales associates on what types of products and product categories are typically purchased together to better recommend these products to in-store shoppers.

Recently, TechStyle announced that they would be taking their media and ad teams in-house. Why did you take the call, what have been the advantages so far, and in what circumstances do you recommend other brands consider the same path?

At TechStyle, we prefer to buy media in-house because it provides more control over the ad buying strategies and outcomes. TechStyle has very robust measurement capabilities that are applied across all channels, which helps our brands buy media more effectively.

Our brands also develop most of the creative in-house and have a very strong creative team coming up with new concepts while incorporating insights from past top performers. By keeping all these processes in-house, we can adapt to change quicker and more effectively spot trends to inform our next big wins.

Learn More: Data-Driven Marketing Trends for the AI Age: Q&A; With Palace Resorts’ VP of Global Marketing, Kathy Halpern

Would you like to share some trends or predictions about the ecommerce and retail fashion space for 2020 and beyond? 

We’ve seen the retail fashion space evolve so much in the last two years, but given the rapid pace of technology innovation, I expect much more change soon.

Beyond advancing omnichannel strategies, in 2020 I predict that messaging platforms will become the core way brands communicate with their customers. I am personally so excited to see the impact machine learning has on ecommerce, from enhancing 1:1 personalization to improving dynamic merchandising.

I feel so lucky to be in this industry right now with so many exciting developments underway that are poised to forever change the way people shop.  

Neha: Thank you, Danielle for sharing your insights on the importance of machine learning for personalization. We hope to talk to you again soon.

About Danielle Boeglin:

Danielle Boeglin is the VP of Data Analytics at TechStyle Fashion Group, the global fashion retailer known for its membership-based digital brands Fabletics, Savage X Fenty, JustFab, ShoeDazzle and FabKids. In her role she facilitates collaboration between the company’s data science teams and marketers to optimize business decisions using customer analytics.

About TechStyle:

TechStyle Fashion Group is a global fashion and lifestyle company founded in 2010 to deliver access, quality and style for unprecedented value. TechStyle Fashion Group uniquely merges advanced technology with the latest fashion trends to offer an entirely new shopping experience to millions of customers worldwide, including five million VIP Members, through a portfolio of apparel, accessories and shoe brands. TechStyle Fashion Group is reimagining the business of fashion through data, personalization and vertical integration to benefit the modern shopper.

What are your thoughts about getting personalization right? Follow us on Twitter, Facebook and LinkedIn to know more about the latest in martech trends across the globe.