American Apparel: driving customer centricity in an omnichannel world

American Apparel’s chief digital officer discussed the future of retail, the importance of delivering value to the consumer, and strategies for an IoT and omnichannel world.

Thoryn Stephens, chief digital officer, American Apparel, was speaking at ClickZ Live Hong Kong where he delivered a keynote presentation on how marketers can drive value to the consumer using data and technology. Here’s how:

How do we define value, how do we measure value and how do we drive value?

Stephens always wanted to be a rock star. Instead he became a molecular biologist. It was this scientific approach to analyzing data that eventually lead him in 2003 to linking a Google Analytics account with Ad Words and Salesforce for his first foray into understanding consumer behavior.

This in turn has led to roles with Silicon Valley startups, a television broadcaster and today, as CDO at American Apparel.

“Being a former scientist, everything I do is based on measurement,” said Stephens.

He believes he receives more funding than any other department heads because he is able to show how everything is measured. He also believes part of his success comes down to owning his own technology team.

According to Stephens, to drive traffic from acquisition to conversion, marketers must:

• Understand the consumer experience
• Optimize through test and learn
• Develop retention strategies

The foundational elements to all of this are technology and data science.

To begin, organizations need to understand where they are at in the data maturation curve.

The four steps of the data maturation curve are:

1. Collecting the right data

2. Reporting on data and driving insights

3. Hypothesis testing (and developing a test and learn approach)

4. Predictive analysis

Each stage on the curve becomes more complex, but simultaneously increases the potential business impact, the further along the curve an (Read more...) reaches.

User states and profiles

Stephens looks at the world in three major dimensions:

• The unknown user
• The anonymous user
• The known user

Each one of those dimensions has a value. Once users have been identified, the aim is to drive the consumer from an unknown or anonymous user, to a known user.

The unknown user

During his time at Fox Broadcasting, Stephens was responsible for building the business’s first data science optimization strategy. At the time, he was working on The Simpsons brand. It had more than 70 million likes (unknown users) on its Facebook page, but by comparison, CRM was small (known users).

The key challenge was to convert these unknown users to known ones. By using a Facebook application, such as a sweepstake, Stephens and his team engaged users and incentivized them to share an email address. With an email address, the unknown user became a known one.

The anonymous user

An anonymous user is a user who might be cookied for example. There is a basic profile around behavior or geo-location, but it’s not known exactly who they are.

Using retargeting, a marketer can start tailoring content to these consumers and drive them into becoming known users. This can be achieved by encouraging them to:

• register for an event
• make a purchase
• download an app (with registration)

The known user

Once a user becomes a known user, all sorts of data can be pulled on them.

“Ultimately you can drive the most value from them not only as a consumer, but now as a brand as well, because you are tailoring the experience to that known individual user,” said Stephens.

By identifying users in this way, they can be better targeted across channels also.

• The unknown user can be targeted with TV and video.
• The anonymous user through retargeting based on site experience with ads.
• The known user with apps, push notifications, in-app messages, SMS and email.

“Ultimately, every single one of these interactions outside the unknown users, I can track and understand and ultimately optimize,” said Stephens.

Driving customer centricity through user-level valuation

According to Stephens, customer centricity focuses on the current and future needs of a select set of customers to maximize long term value for the business.

It comes down to the 80:20 rule, said Stephens.

“You are focusing on the 20% of the customers driving 80% of your revenue by really understanding who those users are,” he said.

Here’s a breakdown of the different forms of customer values:

• RCV – realized customer value: The value of this customer today.
• RLV – remaining lifetime value: Retaining this customer into the future.
• CLV – customer lifetime value – this is the combination of RCV and RLV and is the value of your consumer for an indeterminate period of time. How much are you willing to pay to acquire this customer? For example, if the lifetime value of a customer is $500 maybe you will be prepared to pay $200-$300 to acquire them.

RCV case study: Silicon Valley startup

In a previous role, Stephens worked for a startup, which had raised $40 million, had 2 million Facebook followers and 10 million members.

However the percentage of customers was quite a bit smaller, and the RCV was in the negative tens and thousands of dollars. Stephens wondered how it was possible to have negative value customers.

From analyzing the data, his team found on a user level, many were serial exchange and returners. Through the specific RCV metric the team were able to immediately change the company’s shipping and exchanges policies, driving additional savings to the business.

Clustering analysis case study: Fox Broadcasting

Using a statistical methodology called clustering, Stephens was able to identify visitors to the Fox website that were all alike in a specific set of ways.

First he pulled together one year’s worth of Adobe analytics data and clustered it.

The team was looking for value metrics that drove the business – in this case consumers watching video – (watchaholics) a primary source of monetization.

Using clustering and a specific algorithm called excitation maximization algorithm (EM) they began to see patterns of behavior. Four major types of users began to emerge within the ecosystem.

1. The watchaholics: people with high frequency visits, coming back to the website and driving the video ad views. This group was considered high value consumers on the valuation curve because they were driving additional ad dollars.

2. The casual watchers

3. The international group

4. The passive group

After focusing on the watchaholics, the team looked at their recency and frequency.

“What could we learn about their behavior? We took that insight and then started testing.” The insight was then used to test against a small subset of casual watchers with the goal of driving them to becoming watchaholics.

“We could have stopped there. But we didn’t,” said Stephens.

The data showed the passive users didn’t consume a lot of the video but they did have a high visit count. The team found that after immediately coming to the Fox website, the second page they visited was the schedule.

“They were actually using fox.com as a giant TV guide. This flipped the valuation curve on its head. It allowed us to understand the behavior of the passive users and how to test against them.”

After creating a segment each time a passive user came to the website, the schedule of the user’s favorite show was placed on the homepage.

Stephens’ team then took it a step further with an email collection for an automated alert.

Audience development case study: American Apparel

Facebook pioneered custom audiences a number of years ago. It allows marketers to take four major variables: email, phone number, a device idea or a cookie and feed it into Facebook and target those consumers through a number of dimensions. It’s now also available on other platforms such as Instagram, Twitter and Google’s Customer Match.

In this example, American Apparel wanted to improve cart abandonment rates. “Through Oracle we know when someone ads to the cart. Literally within seconds if it bounces, instead of getting an email, they get a retargeted ad on Facebook. It can be down to the millisecond.”

Stephens said the returns on these social CRM campaigns are 30 times the return on ad spend.

In another example, American Apparel used the data to look for dormant email subscribers – consumers who had not interacted with the website or opened an email over a period of time. These re-engagement campaigns were distributed as emails or via Facebook, targeting them with discounts.

The future of retail

What does the future look like for marketing at American Apparel? Stephens is focused on a number of areas.

Omnichannel

Omnichannel marketing is the intersection of retail and digital. For example, a consumer is walking down the street and as they walk past a shop, they receive a push notification that drives them into the store to a purchase. This omnichannel attribution credits both the digital and retail channels for a sale.

IoT and RFID

RFID is a chip that can be embedded in the tag of every single piece of apparel. At American Apparel, this technology has been deployed globally across 200 stores and 15 million tags. It allows a marketing team to follow and understand global real time inventory levels.

The next step is establishing how it can be used to improve the consumer experience and drive revenue. American Apparel is currently experimenting with mobile devices linking to ads on a billboard or bus stop using NFC chips. If the consumer likes the ad, they tap their phone to start a text bot conversation. The user can ask questions about colors or sizes and then find out if that specific customised product is available in a nearby store.

On demand with Postmates

Recently, American Apparel ran an on demand delivery campaign with Postmates in the United States. Consumers could order a hoodie and have it delivered within 60 minutes.

“Is the consumer adoption there? Not yet but its getting there,” said Stephens. He highlighted the “must have now” culture of millennials and believes on demand delivery is the way of the future.

Takeaways

Stephens’ key takeaways are as follows:

1. Evaluate organizational data maturation

2. Drive the business towards customer centric measurement and metrics

3. Test and learn – segment high CLV users and test hypotheses to move lower value users to higher value segments

“Test and learn everything. If you understand your high value customers, ask what can you learn about them, and then test your low and mid value consumers to drive them to become high.”

4. On demand fulfillment is the future of retail

Small budgets

A big budget is not essential to understanding lifetime values, said Stephens.

“Many models are available for free. So long as you have your transactional records coming out of  your ecommerce engine, or point of sale, you have all the data you need. Recency and frequency might get a little bit dicey, but generally you should have what you need – a time stamp and a transactional record.”

Let's block ads! (Why?)