That Time When Building a Customer Model Taught Us the Difference Between “Active” and “Active” Customers (not a typo)

There are many ways to be smart about your Retention Marketing. You can use control groups to see what really works. You can measure the effectiveness of your campaigns by monetary uplift –and say goodbye to vanity metrics. You can A/B/n test – but instead of the traditional "winner takes all" method, you can go with different winners in different customer segments (of the same campaign, yes).

You can look into the data, stare it in the eyes, then confidently ignore the "it's how we always did it" comments and prioritize existing customers over new ones in peak times such as Black Friday/Cyber Monday or the World Cup – because it's the right thing to do. You can again defy the traditionalists and follow the numbers that show how your regular, recurring campaigns can actually outperform ad-hoc ones, yes, again, even during holidays or other big events.

There's a lot more where that came from – oceans of insights that can help you boost Customer Lifetime Value significantly. But all of it, at the core, stands on smart customer segmentation – the ability to create AND manage more and more customer groups quickly and easily, in realtime, based on sophisticated data streams and calculations, as your business and your CRM Marketing strategy grow and mature in tandem.

Infuse it all with the power of next-gen AI-based marketing technology, and you could make this all a reality (and we happen to know a guy, wink wink).

But even with an industry-leading AI at the heart of our platform – there's no hyper-segmented Retention Marketing without the all-critical Customer Model – custom-built for each new brand by our ultra-experienced Data Science team here at Optimove.

Get to Know: The Customer Model

The Customer Model is built on "states," also called "Lifecycle Stages" (LS), and on the data-driven, machine learning-based predicted migrations between them. At every point in time, every customer can be in one state or another, but never in two at once. Imagine "New customer" or "Churn." You can't be both.

Then, using a robust ML algorithm, we learn how and what makes customers move from one state to another. Is it promotion X or message Y? Is it an email or an SMS? To each customer, their own preferences and journey – are dictated by their personal behavior.

The building of the Customer Model is where we identify the most meaningful LS that each new client we onboard should build its CRM Marketing strategy around. It's also where we define what each LS means precisely. Of course, each one of our clients gets a customer model of its own – because no two customer bases are the same.

For example, the insights our Data Science team extracted from the customer data of Brand A meant their CRM Marketing strategy should include only the basic LS of "New (customer)," "Active," "VIP," and "Churn." It also determined that in this case, a customer remains "Active" only for 30 days since the last purchase.

At the same time, a similar analysis done for Brand B meant that its CRM Marketing strategy should be a little more complex, with the additional LS of "Registered Only," "One Timers," and also "Dormant." In this case, a customer stays "Active" for 90 days since the last purchase.

It's the customer data that dictates the LS, and it's the customer data that also leads to the precise definition of each state. Because, as we said before, not all customer bases were created equal.

Okay, so...

You might wonder why these LS are the core of a hyper-segmented, personalized-at-scale CRM Marketing strategy? And the answer lies in two parts:

1. We enhance the determined LS with customer data-driven Segmentation Layers that create micro-segments on top of the different stages. We define a set of Segmentation Layers for each LS, and these describe key aspects of the customer's behavior. Some segmentation layers are behavioral, while others might be demographic in nature.
   Learn more about micro-segmentation here.
2. We let our Machine Learning map the possibilities for the different migrations between two micro-segments and what is needed to encourage the more desired migration. That's where our Predictive Analytics does its magic.

From here, calculating the future value of each customer and determining the best-next CRM Marketing move for each customer at any touchpoint is "the easy part."

Having a platform that does all that for you in realtime is what allows our clients to basically provide a true 1-to-1 customer journey that is always personalized and maximizing Customer Lifetime Value. And that's why there's no personalized-at-scale without the customer model.

A Different Kind of Active

Recently, our Data Science team ran into an interesting case with an online investment service. Looking into this brand's customer base's behavior, it looked like there wasn't much of a difference between their "New" and "Active" customers. At least, on the surface of things.

Because, if you think about it – if a customer signed up AND made an investment, they are still "active" customers even six months later, even if they did not do anything since – in the sense that they still have money invested through the app, which they can monitor and add to or withdraw every day.

But, in order to build the most valuable Customer Model possible, there is a need to distinguish between New and… non-new because we want to be able to see what the lifecycle of a customer looks like over time.

And so, by digging deeper into the data, our Data Science team decided the following:

1. Because 63% of all SECOND deposits were made within the first 31 days since a customer joined and made their first deposit – the "New Customer" LS will be set for 31 days.

Then, after that period, they will move on to "Active."

1. (unless the customer withdrew their money and closed their portfolio within the first 31 days – in which case after the "New" period ends, they will move to "Churn")
2. But, the "Active" LS is now split into two kinds of "Active":
   1. "Active Investors" – customers who are still investors but did not make a second deposit.
   2. "Active Depositors" – customers who made a second deposit.

Looking into the data, we also saw that among "Active Depositors," if a customer deposited in the past 30 days – there is a 95% chance they will make another one in 6 months. That kind of analysis is where we can start getting more "predictive."

Additionally, about 89% of "Active Depositors" make an additional deposit every 60 days – and 2 in 3 make a deposit every month. Had the customer did not make another deposit within 60 days? They now fall into the 11% that deposit about once or twice a year. Still "Active," but – not all "Active" customers are equal. Not even within the same customer base.

And so, as we said – there are many ways to be smart about your Retention Marketing. And now, it appears that "treating different kinds of Active customers with different CRM Marketing strategies" can be added to the list.