Combat Attrition with Automated Machine Learning (AML)

Combat Attrition with Automated Machine Learning (AML)

For brands across industries, there can be as many reasons for customer churn as there are customers. But whether it’s in retail, banking, insurance, telecom, media, or even hospitality, a poor customer experience is a universal indicator for attrition.

In a Harris Poll sponsored by Redpoint, 37 percent of consumers said that they will not do business with any company that fails to offer a personalized experience. Furthermore, in the PwC Future of Customer Experience Survey, 32 percent of respondents said that they will stop doing business with a brand they loved after one bad experience. (For US customers, 59 percent said they will walk away after “several” bad experiences, 17 percent after just one bad experience.)

A positive customer experience, then, is widely accepted as the first line of defense against customer attrition, especially with price and product having largely been commoditized. The expectation brands must meet is for a seamless CX across an omnichannel journey that spans digital and physical channels. Because the cost of acquiring a new customer exceeds the cost of retaining an existing one (5X more, according to some studies), companies are incented to provide a differentiated experience that treats each customer as an individual.

Much as a superior CX comes down to predicting with a fair degree of accuracy relevant experiences for a single customer, predicting churn is about picking up signals at the customer level. One customer’s reasons for leaving may be the same reasons another customer decides to stay, which makes it important to have a granular understanding of each customer.

A granular understanding, in parallel with common churn indicators that vary by industry, such as closing a bank account, letting licenses lapse for a SaaS company, or a reduction in monthly spend at a retailer, together form a better predictor of churn than signals taken separately. As with delivering a personalized CX, analyzing churn signals starts with data. But with so many reasons why a customer might leave, and an even greater number of permutations, accurately predicting churn at the individual customer level at scale depends on automated machine learning (AML).

AML Reveals Customers’ Intentions – At Scale

While it’s rightly accepted that predicting the behavior of millions, tens of millions or even hundreds of millions of customers is beyond human ability, it’s worthwhile to look at one alternative that many companies still rely on to predict churn – and why it falls short for today’s omnichannel customer journeys.

That is, many companies still rely on the common churn predictors – the closed bank account, the reduction in monthly spend, the negative NPS survey, notes from a call center complaint, etc. A good start, perhaps, but as a standalone each fails to account for an entirety of a customer’s behaviors along a customer journey – an analysis of which might lessen the importance of one churn indicator as it relates to the journey.

A retail example might be a retailer automating a retention outreach if average monthly spend dips below a certain percentage. Yet without knowing more about a customer, the retailer runs the risk of introducing friction into the customer’s journey. Perhaps the customer has recently lost a job. Or maybe the customer made online purchases from the retailer using a household member’s credit card. From the customer’s perspective, a standard retention email will seem insensitive.

Self-Training Models Keep up with the Customer

AML reduces the possibility of introducing friction due to a mistaken interpretation of churn signals with in-line, self-training and continually optimized models that do not require any human intervention. Code-free models and automated algorithmic optimization mean that a model set for a specific business metric – in this case, retention – will run 24/7 testing various permutations to determine the optimal retention action to take for each customer, based on each customer’s churn signals.

Organizations can run fleets of models simultaneously, with an evolutionary programming capability determining not only winners and losers, but a winner optimized against the chosen metric for a specific moment in time. A closed savings account may queue a default action barring competing behaviors, but if the owner of the closed account immediately follows it up with a call to customer service to open a new account with a better interest rate, a model will incorporate changes to the data into subsequent production models.

New data may be at the customer level, but a change that influences churn rates may also be at the enterprise level. A telecom company with an increase in the number of internet outages may see a significant uptick in churn signals, which self-training models will account for in an updated prediction calculus.

AML guarantees that a brand is always in sync with customer intent, and is analyzing churn signals that are precisely aligned with the customer at an exact moment of a customer journey. Because the Redpoint rgOne platform combines AML with a real-time decisioning engine, a brand is empowered to react to churn signals in real time – or at the exact moment an action will optimally drive the intended behavior. Intelligent orchestration capabilities enable brands to react to churn signals in real time at scale, for any number of customers, and for any combination or sequence of behaviors that may predict a single customer’s propensity to churn.

In the Harris Poll referenced above, 63 percent of customers said that personalization is now a standard expectation. Asked to define what it means to them, 43 percent said it was a brand recognizing them as the same customer across every channel. A brand that does this well is already safeguarding against churn, but to deliver a personalized omnichannel CX without giving churn its proper due is to risk alienating an existing customer who might otherwise receive a seamless – and frictionless – experience throughout a journey. A perfectly timed, AML-optimized retention outreach that stops churn in its tracks is another arrow in a marketer’s quiver to deliver a superior experience.

Related Content

Retention Marketing for the Digital-First Customer: Make an Impression with a Golden Record

To Optimize Revenue Growth, Tune in to Customer Lifetime Value

Algorithmic Optimization and the Magic of AML

Be in-the-know with all the latest customer engagement, data management, and Redpoint Global news by following us on LinkedInTwitter, and Facebook.

Get Started on Getting Ahead

Schedule a conversation and learn how Redpoint can put your goals within reach.

Get Started on Getting Ahead

Schedule a conversation and learn how Redpoint can put your goals within reach.

Email Your Molecule
Do Not Sell

Submit the form below to set a "Do Not Sell" preference for your user within our persistent customer records.

Meet with a Redpoint Partner
Hello there!

Please fill out the form below and we will reach out to you.

Open

Your Unique rgOne™ Solution

Click below to create a personalized Molecule to meet your specific goals.

Create Your Molecule