The Real Story on Composable CDPs

The term “Composable CDPs” has gained traction amongst MarTech vendors. Proponents of this term promise a new, almost revolutionary approach to managing and activating customer data.

But is it new or revolutionary? And by the way, what does “composable” really mean? Let’s dive deeper.

Composability in Context

Composability, as it relates to tech stacks or architectures, is not a new phenomenon. It has roots in system design and software engineering, and predates the current CDP buzz. Composable architectures are based on principles of modularity, interoperability, scalability, and flexibility. In other words, systems are designed with interchangeable components, so they can be combined in various configurations in a scalable way. At RSG we typically advocate for composable MarTech stacks.

But what about individual platforms? In more recent times, we have seen composable architectures in the context of CMS and eCommerce systems as opposed to traditional monolithic systems. Composable architectures allow organizations to select best-of-breed solutions for specific functionalities and integrate them into a cohesive ecosystem. For example, you may no longer want Personalization and Analytics services from your CMS.? Traditional monolithic systems, however robust, usually lack the flexibility to address evolving architectures.

On the other hand, the definition of "composable" propagated by some customer data platform? vendors appears to be narrowly focused. They essentially conflate composability with the mere use of a Data Warehouse (DWH) for storage. This perspective falls significantly short of what enterprises require from a truly composable ecosystem, and on their face can turn out to prove less composable than meets the eye.?

Composability: A Spectrum, Not a Feature

Let’s start here: In an ideal scenario, a composable CDP should enable organizations to seamlessly build a robust data management ecosystem by integrating best-of-breed tools, offering flexibility and scalability beyond just data storage solutions.

Composability is a pattern that requires careful strategic consideration in the context of overall business strategy and technology integration. Yet, in stacks as well as discrete platforms, composability always represents an evolving continuum. In other words, composability varies in degree rather than existing as a binary feature.?

Stalking Horse for Reverse-ETL

Many vendors that today champion composable CDPs originally started with a very narrow focus, mostly offering reverse-ETL (Extract, Transform, Load) services. Today the concept of a composable CDP too often serves as a strategic facade to mask their offerings' limited features.

I have always believed reverse-ETL is a feature or product – and not a platform – and it was only a matter of time before activation platforms started embedding reverse-ETL capabilities, thus doing away with the need of dedicated reverse-ETL tools. In response, reverse-ETL vendors started building adjacent capabilities. “Composable CDP'' became a convenient umbrella under which to expand their feature sets.

Misplaced Enthusiasm Among CDP Vendors

Marketing messaging by composable CDP vendors is often linked to DWH providers. In fact, many times, it is not clear if that long blog post by a composable CDP vendor’s marketing team is about their own capabilities or about a newly launched DWH feature. These blurred lines do little to clarify those CDP vendors’ value proposition.?

The term Composable CDPs also buckets broadly-featured CDPs and narrowly-focused tools such as reverse-ETL tools together. It’s intriguing to note that some comprehensive CDP vendors are okay with this. But this actually muddies the water more by making it difficult for enterprise MarTech and DataOps leaders to distinguish between different types of tools.

By the way, can you imagine what happens when DWH vendors start offering their own CDPs, either by building capabilities or acquiring vendors? After all, they have acquired vendors in adjacent spaces such as Data Clean Rooms (DCRs). When that happens, will composable CDP vendors still want to ride the DWH wave?

Targeting the Wrong Enemy

Reverse ETL vendors have frequently positioned traditional CDPs as their primary competitors. This is misguided. I’d argue that the real challenge for end users in the industry is not whether to use a composable CDP vs a traditional or packaged CDP.?

The real challenge is how to address poor architectural practices and monolithic suites that hinder agility and flexibility. By focussing on the wrong problem, these vendors have missed the opportunity to address core issues facing enterprises.?

Are Composable CDPs Really Less Expensive?

At Real Story Group this year we’ve run several competitive selections, involving both kinds of CDPs. And composable CDPs did not prove less costly, even on face value. But when we start peeling the layers, there’s more.?

A key argument made in favor of composable CDPs is the potential for storage cost savings due to data residing in DWHs. First of all, remember, storage is comparatively cheap and it keeps going down.?

So the followup argument is that you can handover all your expensive queries to DWH and keep CDP lighter in terms of compute (and hence lower costs). But the reality is that compute costs merely get shifted upstream, impacting overall efficiency and cost-effectiveness. In fact, I would argue that it can actually increase your overall compute costs because now you have to keep your DWH cluster always active/on or at least more than before, and that has a huge impact on your costs.

The Assumption of Universal DWH Adoption

“Everyone has a Data Warehouse.” Maybe. But can everyone in the enterprise use the same DWH? Should they?

The assumption that every company has or should use a DWH as the center of their marketing technology stack is flawed.? There can be several reasons but they can usually be clubbed into these two:

• Organization may not have a DWH
• Organization may not want to use an existing DWH for marketing

Various reasons, ranging from governance concerns to strategic priorities, may lead an organization to decide against integrating their marketing efforts with a DWH. This reality challenges the one-size-fits-all approach advocated by composable CDP vendors.

Single Source of Truth…Not!

Another benefit touted by proponents of composable CDPs is a "zero-copy" architecture. They claim that data does not need to be replicated from the DWH, and therefore there is a single source of truth for your customer data. And that’s good because having a single copy of data helps you with data privacy and security in addition to reduced storage, plus other benefits.

That’s a great idea, in theory. In real life, a single source of truth doesn’t work that well. We have seen this in the past with promises made by Enterprise Content Management (ECM) systems. That didn’t work out so well. It oversimplifies the complexities of data management.?

In practice, data often needs to be replicated or transformed for various activation platforms upstream. This does replicate customer data, thus undermining the claim of a single source of truth and highlighting the limitations of this approach. In the best case scenario, it’s just not getting copied within the CDP but subset of it goes everywhere else.? Ironically, with this approach, you can end up copying more data to engagement platforms, forcing more complicated data management and segmentation there, rather than a CDP.??

And need I remind you that this approach can also complicated your security and data privacy / compliance regime.

The Myth of Escape from Vendor Lock-in

Composable CDP vendors also advocate that their solutions are a means to avoid the pitfalls of vendor lock-in associated with traditional CDPs. Their argument is that with traditional CDPs, your data gets locked into proprietary repositories. But they inadvertently introduce another form of lock-in with Data Warehouse (DWH) providers. This scenario raises critical questions about the true freedom from vendor dependencies and highlights the need for a more nuanced discussion on vendor lock-in within the data platform ecosystem.

What You Should Do?

At RSG we don’t advise vendors. But I’ll make a small exception here.

Alternatives are good. The idea of using composable architectures is good too. But bad marketing isn’t a good idea for anyone. It’s bad for you, it’s even worse for your customers.

Now some advice for you, the enterprise MarTech leaders.

Deploying a composable MarTech stack makes very good sense However, don’t just go by vendor-propagated definitions of what composability entails. Define your own requirements of composability. And if that means you want to leverage your DWH for customer data activation, that’s fine, as long as you understand all the (potentially unpleasant) implications. Don’t just limit composability to that, though.?

You should critically examine the claims of MarTech vendors against your real-life use cases and challenges. Look at the nuances. For example, if a vendor suggests using their tools will reduce your costs, ask for independent studies to verify that. And then adapt those studies for your specific context to see if you will actually benefit.?

Of course, RSG can help.? Ping me for details.

Meanwhile, you can check out how different CDPs fare in this marketplace using our Real Time Decision Tools or contact us if you’d like help selecting the right CDP for your needs.