There’s no question that collecting and sending customer data has become a top area of emphasis for organizations globally — especially when it comes to the necessity for that data to be real time.

That’s where RudderStack Inc., described as a “warehouse-native” customer data platform, comes into play, with the goal of enabling companies to collect, unify and activate its customers’ data.

“Warehouse-native, what that means is instead of storing data, RudderStack actually turns a customer’s data warehouse, or data lake, into a customer data platform and handles all of the collection, the modeling, [or] unification and the activation of that data,” said Eric Omwega (pictured), vice president of marketing and operations at RudderStack.

Omwega spoke with theCUBE industry analyst Lisa Martin at the “Analytics and Cost Optimization” AWS Startup Showcase, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed how the company’s platform enables real-time event data capture and real-time and batch use cases, ensuring the optimal use of data for various purposes, including churn prediction and other machine learning or AI applications. (* Disclosure below.)

Enabling customers to deal with event data

When RudderStack first started, it built its Event Stream pipeline, which collects customer data from one’s website or app so that each interaction that a customer has with a web or app property is an event that’s being created, according to Omwega. After collection, instead of RudderStack storing the data — which is what traditional customer data platforms do — the company actually sends that data, in real time, to downstream destinations. For example, if one wanted to activate real-time use cases, that data is collected and sent in real time to downstream destinations, which could be an email marketing platform or a Facebook ads platform.

“In addition to that, we also have the ability to send that data into your warehouse or data lake, whether it’s AWS S3 or Redshift,” Omwega said. “We expose that data for you to be able to model that data to build, whether it’s machine learning models or whether it’s artificial intelligence use cases.”

Organizations can do all of that modeling in their warehouses and then use the company’s reverse extract, transform, load pipeline to activate the data as well. There are both real-time and batch use cases involved — real time, of course, being done with very limited lag, Omwega pointed out.

There are also batch use cases, as when one is sending data to a data warehouse, they may not want to do that in real time. Users may also want to activate real-time use cases, including for personalization.

“When a customer is interacting with the app or website, already having pre-computed features, a feature could be something like a customer’s likelihood of churning,” Omwega said. “If you have a churn score and you have that customer who’s been flagged as having a high likelihood to churn, you could actually send a notification in real time. Maybe it’s a 20% off coupon in real time, as the customer is engaging with the app. That’s an example of a real-time use case that RudderStack can activate.”

The value of personalization

Personalization most often drives a 10 to 15% revenue lift, according to a 2021 study from McKinsey and Co. The most basic level of what RudderStack provides is what was previously discussed — the company’s real-time activation using its Event Stream pipeline, sending it to downstream destinations, according to Omwega.

“But then, you start to think about the more advanced customers get. For example, they want to calculate ML features in real time,” he said. “There’s a number of different steps to that.”

The first step is data collection, which involves a number of different data sources, including software-as-a-service data, but also customer service platforms, such as Zendesk Inc.

“You have two types of data: behavioral data and relational data. You collect that into your data lake or your warehouse. That’s step one, data collection,” Omwega said. “The second one is unification. You need to resolve the identity of that data so that you can say, ‘This specific customer, this is their behavioral data.’”

Once that customer profile is in place, a company can now calculate and enable features.

“A feature could be revenue over the last six months or the last time a user logged in. You calculate those values. Now, in your warehouse, you have a user, the entire customer journey from behavioral data and additional features that could either be deterministic or probabilistic,” Omwega said. “You can unlock true personalization, where you are using all of the data that you have on the customer to deliver the experience that they need to make that next purchase.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of the “Analytics and Cost Optimization” AWS Startup Showcase event:

(* Disclosure: RudderStack Inc. sponsored this segment of theCUBE. Neither Rudderstack nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE