Composable CDP or SaaS? Flexibility is key
Composable has been the big buzzword in the world of CDP for a while now. But like so many tech buzzwords, it’s hard to pick the hype and the reality apart to work out whether it’s a solution that’s going to suit your needs.
So let’s start with a definition. This one comes from our friends over at Hightouch. According to them, a composable CDP is “one that enables you to use any data in your organisation to power marketing use-cases like audience management, journey orchestration, personalisation, and data activation directly from your existing data infrastructure.”
In other words, you as a user are deploying the CDP on your computer infrastructure. With SaaS you’re moving your data into your service provider’s environment. Composable means taking the code to the data, not the other way around.
This has significant advantages, particularly around privacy and security. Your data never leaves your environment. That’s what people mean when they talk about a zero-copy approach.
But while some people would have you believe composable is the silver bullet for CDPs, it also has its drawbacks. One is that you’re not always in control of your costs. Composable means the software you’re running won’t necessarily have been optimised for your environment. In contrast, SaaS is running on the vendor’s own cloud, so they’ll have optimised the hell out of it for their own cost reduction purposes.
You also have to be certain you have the appropriate monitoring and controls in place if you’re taking a composable CDP approach. With SaaS, vendors monitor usage very closely, again because it affects their costs. Composable puts the onus on you to make sure your users aren’t just going full-on burning through compute and data usage. You have to be sure your DevOps is running correctly. And not only can composable add to your cost burden, it can also be difficult to understand and evaluate that ahead of deployment.
Even the privacy benefits can turn into a double-edged sword. Zero-copy is great for batch use-cases and tight data governance, but most real-time use-cases require a fast, accessible copy of your data, which means you either compromise on performance, with slower response times and limited real-time functionality, or quietly reintroduce copies, undermining the original privacy benefit.
One size doesn’t fit all
None of this is to play down the very real benefits of a composable approach. It’s just to point out that, as is so often the case with software, one size doesn’t fit all. You always need to balance your goals, your use-cases and your corporate infrastructure requirements. Sometimes the answer will be a composable CDP, sometimes it’ll be SaaS, and sometimes it’ll be some hybrid of the two. There’s no reason why you couldn’t run a model where certain elements are composable, but there’s a copy of data taken for certain use-cases as well.
Too often it feels like the question being asked is whether one approach or another is better. In the end, the only question that matters is which one is better for your business.
That’s why you need to work with someone who’s a partner, rather than just a vendor. You need someone who understands all the ramifications of the different approaches, and can guide you to the solution that works best for you. Find someone you trust, someone you can have open discussions with, and that way you can go into this with your eyes wide open.
