Analytics In Action: Implementing a Customer-Centric Approach At Scale
By Professor Eric T. Bradlow (@ebradlow) (GBK Collective and the Wharton School)
Earlier this Fall, I had the pleasure of co-moderating “Analytics in Action: It’s Still All About the Customer”, an event co-hosted by Wharton Customer Analytics and Teradata. In part two of my blog series, I explore how to implement a customer centric approach at scale with comments from my colleagues at the Wharton School, Professors Katy Milkman and Kartik Hosanagar.
As brands look to become customer centric, there’s an incredible opportunity in applying data and analytics to optimize decision making, solve business problems, drive stronger growth and increase CLV. But there are a number of variables and challenges to consider when it comes to doing this at scale.
As I explored in my previous post, marketers need to consider both cross-person heterogeneity and individual customer preferences when predicting customer behavior. In other words, depending on the context, an individual customer will have different needs or an entirely different persona.
As the customer journey continues to evolve, brands and marketers have to constantly question their assumptions and seek out better data to fully understand how to deliver value in the moment - based on where customers are and what they’re doing.
Understanding the context and barriers to change
“There is no single answer or silver bullet when it comes to understanding customer behavior through data science,” shares Milkman, who is also the author of the best selling book, How To Change. “In addition to studying within-person heterogeneity by context, you also need to look at what the specific barriers to change are in that environment.”
“There is no single answer or silver bullet when it comes to understanding customer behavior through data science.” - Wharton Professor Katy Milkman
“For example, if people aren't visiting the gym regularly, it might be because they find the gym miserable. It might be because they haven't figured out a way to make time for it. And if you try to use the same solution to all those different kinds of problems, you actually don't end up doing very well.”
The key point here is a customer will have different barriers to change depending on the specific context or timing. And so one blanket approach, even for a given individual, might not work.
The Fresh Start Effect
Another factor brands need to consider is what Milkman calls The Fresh Start Effect.
“As people, we think about our lives through chapters,” shares Milkman. “For example, the years you were in college, the different jobs you’ve held and the places you’ve lived. And with each change, you feel like you’re turning the proverbial page; you have a fresh start and with that comes a new mindset, a willingness to try something new. But fresh starts can also arise at the start of a new week, new month or following a holiday such as New Years, your birthday, or the beginning of Spring.”
For marketers who are looking to improve customer lifetime value or reduce churn, The Fresh Start Effect brings up an interesting concept. Are people more or less loyal because they see it as a fresh start on certain days? And are people actually more likely to buy a product or service or take action on days they see as fresh starts?
“For marketers who are looking to improve customer lifetime value or reduce churn, The Fresh Start Effect brings up an interesting concept.”
Milkman's research and the work shows that by strategically identifying and overcoming common barriers to change and focusing on the right fresh start date, timing and approach, brands are more likely to drive growth or move customers to action.
Algorithmic personalization at scale
In the past, understanding within-person heterogeneity was more difficult for marketers because we didn’t know where customers were. Now by applying better data from a variety of sources, brands can better predict specific outcomes with customers based on location, timing, context or other individual preferences by channel.
“There are a lot of pieces to the puzzle to consider when it comes to algorithmic personalization at scale,” shares Professor Hosanagar. “While most companies leverage some form of machine learning or AI to customize the content and experiences for customers, many of these efforts only account for cross-person heterogeneity. When you look at the personalization by Amazon, Google, Netflix or other leaders in the space, they’re predicting content and experiences down to the individual level. They have embraced the fact that there is both cross-person and within-customer heterogeneity.”
“There are a lot of pieces to the puzzle to consider when it comes to algorithmic personalization at scale.” - Wharton Professor Kartik Hosanagar
As individual customers provide signals and information about their needs and intentions through activities, brands can capture that data to respond with relevant and timely content and experiences based on those triggers. Done right, personalization at scale can enhance customer experiences, increase engagement and improve CLV.
But there are also a number of challenges and risks brands need to consider when it comes to automation and personalizing experiences at scale.
“Two of the biggest and most obvious risks brands need to consider are data quality and privacy,” notes Hosanagar. “Another risk associated with automated decisions is bias. When you have a machine learning system making decisions based on the way customers have behaved in the past and automate based on that data alone, you are introducing bias.”
Why this all matters
Once you have a well scoped overview of the business problem you’re solving for or attempting to predict with customers, the next step is to collect meaningful data by asking the right questions up front. What data is truly needed to improve CX and customer lifetime value over time? What data is needed to better understand both cross-person and within-person heterogeneity?
Another key to success is maintaining a test and learn mindset. To separate the signal from noise, brands need to run constant experiments to learn more about their customers to see what actually works and what doesn’t. The iterative adaptation based on these experiments builds momentum.
This is part two of a two-part blog series. In part one, I discuss how to create a customer-centric and data-driven culture with insights from GBK President Jeremy Korst, Teradata CMO Martyn Etherington.