What CDPs got right–and wrong–about behavioral events

Today we launched Event Streaming, which makes it easy to forward events in real-time directly to your downstream destinations. This is the type of feature that people are used to in traditional Customer Data Platforms (CDPs) and customer data infrastructure platforms (CDIs)– so why did we build our own version of it?

CDPs did many things right with their event streaming solutions– which helped most of them, including Segment, reach widespread adoption. Indeed, some of our own customers who are leaving CDPs behind have explicitly asked for CDP-style event streaming with Hightouch.

However, CDP event collection isn’t perfect. We see an opportunity to both improve traditional event streaming and ultimately power richer data activation in real-time through the warehouse with Streaming Reverse ETL.

In this blog, we’ll reflect on the things that CDPs have done for event collection that have been really effective– and then the areas where they fall short, and where our warehouse-native solutions can outperform the status quo.

Things CDPs got right about event collection

Enterprise CDPs like Segment established a strong baseline industry standard for behavioral event collection, which has shaped how every new entrant builds their products.

Easy-to-implement SDKs. Segment’s event collection can be implemented on websites or in apps with simple software development kits (SDKs). Many solutions, including Rudderstack and Hightouch Events, forked event collection SDKs from Segment’s open-source solution. As a result, most event collection solutions operate similarly, even if they aren’t literally forked from Segment. This makes it easy for users to set up event collection and switch between vendors as they see fit.

Data Contracts. Along with easy-to-implement SDKs that collect behavioral events, most CDPs offer other helpful features to make implementing and managing event collection simple. CDPs have offered (or mandated) standard specs for how events should be collected for different use cases. While these specs can be restrictive, their very existence has shaped general consensus on best practices across the industry. Similarly, most CDPs offer some version of what we call Data Contracts, which allow teams to define expected event data formats, and catch errors and exceptions that don’t conform to norms.

Streaming. Event streaming, which passes behavioral events in real-time to downstream destinations, has become a widely adopted solution for keeping many tools up to date with data at low latencies. Event streaming typically relies on a simple, code-free UI to define which events go to each destination, and how fields should map from your point of collection into each tool. The net result is that engineers can avoid the pain of instrumenting redundant SDKs or manually translating data to match requirements for each downstream system, while marketers benefit from real-time analytics or timely marketing triggered by user behavior.

Things CDPs got wrong about event collection

Despite their industry-defining approach to event collection, traditional CDPs have made critical errors that have left customers frustrated and looking for alternatives.

Too rigid. Most CDPs are very rigid in their expectations for data schema and formats. If you can’t exactly match their event spec, you can’t use their platform. This simply doesn’t work for many businesses with custom needs– and is one of the main reasons that customers like Minno leave traditional CDPs for Hightouch Events.

Hostile pricing. CDP event collection pricing is also a serious burden for buyers. CDPs typically price event collection on hard-to-measure (and even harder-to-manage) “monthly tracked users” (MTUs). Furthermore, CDPs tend to gate essential features like data contracts behind higher-tier payment plans, meaning that you have to pay more to ensure you get quality data.

Poor warehouse connectivity. Traditional CDPs are also inadequately equipped for the rise of the data warehouse. Data that goes through warehouse modeling is richer than raw event stream data: it can be deduplicated, unified into distinct customer profiles, enriched with machine learning models, and more. Traditional CDPs rely on their own data silo separate from the warehouse and miss out on many of these benefits. A better solution for rich low-latency event data is to stream events directly into the warehouse with Hightouch Events, and then rapidly model and activate that data with Streaming Reverse ETL

We’ve made it easy to replace your CDP’s event collection

We’re helping customers leave their traditional CDPs behind, in favor of a Composable CDP built on the data warehouse that achieves better outcomes at a lower cost. Our Event Streaming feature will make this migration, which we typically implement in just a few weeks (or even minutes), even easier: companies can choose to re-create their CDP event streams one-to-one in Hightouch.

We’d love to hear from you. If you’re using a CDP for event collection, what’s going well for you– and what’s not? You can reach out to me at tejas@hightouch.com, or book a demo with our solutions engineers to try out Hightouch Events for yourself.