Traditional marketing is an optimization process. Take a fixed set of resources like budget, people, and time—and optimize for a metric (like conversation rate, time on site, etc.). One popular traditional method of optimization is segmentation, in which you carve up the population into bite-size chunks and optimize the deployment of resources toward that a high-value segment. This is in stark contrast to personalization, which strives to establish a more meaningful 1-to-1 relationship between a brand and its customers.


A segmentation-based approach means iteratively identifying and targeting the largest opportunity. Marketers identify a segment (subset of people identified by a set of parameters) within a channel (mobile, web, display, etc.), and introduces that segment to an experience tailored to what the marketer thinks they’re likely to respond to. Then they observe and assess the impact of the new experience and decide if they should expand it to a larger audience. When you’re talking about single-channel, single-session segmentation, this is a relatively simple task.

But segment-based testing is NOT true personalization. As you iterate, opportunities get smaller and yield diminishing returns. You eventually hit a plateau, where marketing spend is greater than the return. Creative content becomes a challenge and intuition may not deliver beneficial business results. Quite simply: you run out of worthwhile opportunities.


Compare this to true 1-to-1 personalization, where the opportunities are limitless. Consider the following definition of personalization:

Delivering an individualized experience for each customer, in every moment, across every touchpoint, based on everything you know about them.

This approach requires more than just some clever JavaScript on a page. It requires an omniscient, omnipresent platform. More specifically, the platform needs to be capable of intelligent data assimilation and bi-directional data/event exchange in real-time. This means deciding how to react to events to capitalize on opportunities they represent in the most optimal way possible.

back-in-stockTake someone who searches for a product and sees that it’s out of stock. This is a key data point that the platform must store and index for later retrieval. Then when inventory is replenished, the platform can link the inventory update to that customer’s past actions. It can act on that knowledge and reach out to the customer through a channel not controlled by the brand, like a push notification.

This was the vision on which Monetate was founded, and we built our platform to accommodate these experiences. Our decision engine is at the heart of a real-time feedback loop through which we ingest data and events, react to those events, observe the system, and adjust the machine-learning models that underpin the decisions. This is done via models, which are capable of making nuanced decisions while they observe and predict performance along many different dimensions. This allows for individualized decisions and personalization not possible with rigid rule-driven segmentation, which results in better experiences for more people.

True ecommerce personalization should incorporate three types of customer data:

  • First-party brand data like product catalogs and loyalty program information
  • Events such as customer service calls, travel information, or prescription fulfillment
  • Third-party data like weather and census data.

You then need to connect those data points so you can convert events into opportunities. With a personalization platform like Monetate, you can improve key performance indicators with an automated solution that learns and adapts based on the observations and real-time analysis of customer behavior.

To learn more about data-driven personalization, and for tons of real-world case studies, check out the L2 Personalization Report.