Playback speed
Share post
Share post at current time

CDP Rapid Fire - Round 2

Concluding the Composable vs Packaged CDP debate.

This is a continuation of Round 1 — for full context, do check it out if you haven’t already.

Welcome to Round 2 of the CDP Rapid Fire!

In this round, our host, Glenn Vanderlinden asked the guests follow-up questions based on their responses to the statements from round 1.

This one is packed with too much good advice and too many laughs, leaving no reason to miss it.

In fact, there was so much goodness in this episode that I had to cut it short. In the coming weeks, I’ll release the rest as snippets on LinkedIn (connect with me if you haven’t already).

P.S. If you’re a recent subscriber and are wondering what’s with CDPs being all the rage, please have a quick look at our campaign “Let's End The CDP Battle”.

Listen to the full episode or look for “data beats” wherever you get your podcasts!


I’d like to thank our esteemed guests who made the CDP Rapid Fire happen:

Watch the episode above or on LinkedIn, and if you prefer to read, well, keep going...

Four out of the five guests responded with “it depends” to the statement, “Composable CDPs are better suited for engineering-driven organizations.

Glenn’s question for Matthew and Jacques:

What’s the skillset needed to run different types of CDPs — Composable versus Packaged?


The decision usually rests upon the composition of the company and the type of resources they have availableengineering or non-engineering. A group of engineers will take the Composable CDP route, whereas a group of marketeers and salespeople will take the Packaged route.


When you implement a CDP, you need engineering support — this is not something that a marketer or salesperson can just put into place, especially if you want to deploy proper tracking.

In terms of the technical resource available, a (Packaged) CDP is probably less complex because you don't need to stand up a whole piece of data infrastructure.

Packaged CDPs are more suitable for organizations where Identity Resolution is a must-have.

Surprised by the responses to this statement, Glenn asked Boris:

The Composable CDP usually doesn't come with a unification engine — an identity resolution solution that enables a “Customer 360”.

How do customers with complex identity resolution use cases typically approach that problem?


You can't have a Customer 360 with only a single set of inputs — "we know everything customers did on our website and our shopping cart but we don't know what they did in the payment system” — that's the first-level problem where one has to stitch those things together.

And the way our customers do it is via some human, not necessarily that technical but someone who has some amount of understanding of the context inherent in their company.

Jacques, I'm sure you've done things like this that you may have not have called entity matching or identity resolution — this user is associated with this company or that workspace and so on.

And then one has to work with the underlying data model which is somewhat complex.

A human (user) doesn't necessarily just belong to one organization — they might belong to many or that human may actually have many versions of themselves across distinct workspaces.

B2B has this all over the place and I would argue there is no one tool that fits all possible data models. 💯

Glenn asked David:

Do you think going forward, there’s room for Packaged CDPs to offer pieces of their product as standalone solutions?


Our definition of CDP explicitly excludes having identity resolution as a required capability.

We don’t require identity resolution capability in a (Packaged) CDP. The CDP has to make it possible but it doesn’t have to have its own (ID resolution) technology.

Will CDP vendors refactor and sell components separately?

We see a little of that now but I don't really see that as a great business model for the CDP vendors. But technically, if they've factored it out properly and they've made it available as a service in a technical sense — that which some of them have and more of them are doing — it’s doable.

Glenn’s question for Matthew and Jacques:

What's the typical implementation time of a Packaged CDP that you've seen?


I mean, it really depends. I hate to keep saying this.

Depends on the resource you have at hand — the appetite within the company to actually implement this stuff. The thing I've seen a lot is this:

A marketing team will buy a CDP and then kick it over to engineering which hasn’t planned to do the actual instrumentation work.

Therefore, implementation time could be a week or it could be a year.


It depends on the number of sources you want to collect data from, the number of destinations, the technical resources available to implement the solution, and the support from stakeholders and higher management to push the more difficult changes through. 💯

When I do CDP implementation, I usually work with these companies just for the base implementation from anywhere between three to four months.

Question from an audience member, Lee Hammond:

Now that many (Packaged) CDP vendors have rolled out deeper integrations with the warehouse, what are the key differences in taking a composable approach using those tools versus something like Census?

Michael (MK):

This is a bit of a loaded question. It all depends on what you're trying to solve for, right? What's the use case? How do you define value?

The feature and functionality thing that started to dominate much of the conversation in the early days of the “reverse ETL is gonna kill the CDP”, which then became “Composable versus Packaged”, which we're now in — it's all kind of nonsense.

It's what are you trying to accomplish and what's gonna be the best path to get there. The rest of it is nonsense.


First, I'll +1 MK in terms of who's killing what — I hope I never try to frame anything like that in my life.

Composability in software is probably the most valuable thing that helps people build scalable software — it's the thing that allows you to deliver more software with the same number of humans as not everyone has to understand every piece of the system.

More interoperability is good for humans. Period.

Therefore, I'll give it to MK. We should not be thinking about, "You must pick this or you must pick that." But software being interoperable is just unequivocally better for the end user. 💯

Share your thoughts on LinkedIn

Closing thoughts and what’s next

Conducting the live roundtable and learning about the diverse perspectives of these amazing individuals has been, to say the least, an enriching experience for me.

Have the conversations helped clarify my own thinking on the CDP debate? Absolutely!

But have I ever had more fun creating content? Nope, definitely not.

The number of laughs throughout this series is pretty freakin’ high — it’s so much better when people are able to have a few laughs while engaging in discourse about something they’re passionate about.

Moreover, the goal behind this campaign was for folks from both camps to find some common ground, and I believe we did find a fair bit of common ground.

If you’ve followed along, I hope that you now have a better understanding of how the two approaches — Composable and Packaged — differ not only in theory but also in practice.

As mentioned at the start, I decided to cut this episode short and intend to release snippets on LinkedIn where the guests tackle the following questions from the live audience:

  • For CDP implementation projects to be successful using either approach, what prep work do marketing and engineering teams need to do in advance?

  • Can a CDP be successful without leveraging behavioral data?

  • How do you orchestrate experiences downstream across multiple channels using the composable approach?

  • How do you minimize data redundancy if you want to enable near-real-time workflows?

I’d like to conclude this campaign and move on to other important data topics related to People, Privacy, and Personalization.

I will definitely make that happen, but I also want you to know that I am fiercely passionate about the CDP category — have been so for 3+ years. And while I intend to dial down on CDP content, with the space currently brimming with activity, it’s very hard for me to just leave it alone.

Moreover, you gotta give your audience what they ask for, don’t you?

This brings us to the end of the campaign but if you’d like to keep learning about CDPs, check out the CDP Beats series:

databeats newsletter 🥁
databeats newsletter 🥁
Arpit Choudhury of databeats 🥁