At the risk of showing my age, let me take you back to the early 2010s. Marketers were stuck in a digital purgatory: data everywhere, insight nowhere. Every new campaign needed IT involvement, every personalisation dream died somewhere in the dev backlog, and data silos were as common as an “F-3 error” message on the office printer.

Enter the packaged CDP. A knight in shining armour, trotting up on his powerful steed.

It promised freedom. A single place to collect, unify and activate customer data—without begging IT for help. And for a very rare, and particularly persistent (read well-funded) few. These platforms were a genuine leap forward, empowering marketing teams to stitch together fragmented customer journeys and trigger comms with a level of autonomy that felt revolutionary.

But revolutions have a habit of eating their own children.

As these CDPs grew in ambition, so too did their complexity. What started as a marketer-friendly solution morphed into sprawling, opaque ecosystems requiring just as much IT support—only now layered with vendor workflows, rigid schemas, over-cooked identity resolution, “harmonisation” projects and integration gymnastics worthy of Cirque du Soleil. They became the very molasses they were meant to replace.

Meanwhile, the rest of the data world kept evolving. Warehouses like Snowflake and BigQuery went supernova. ELT pipelines matured. Real-time streaming got real.

And yet—many brands are still duplicating the same event data across five different platforms, just to track performance, segment audiences, and fire off campaigns. One copy to the analytics tool. Another to the CDP. One more to the product team. A backup in the data lake. And a sacrificial offering to the ABM platform just in case.

It’s expensive. It’s error-prone. And in today’s tech landscape, it’s kind of bonkers.

Composable CDPs expose just how unnecessary this has all become.

Instead of moving copies of data between platforms, they let your existing systems talk to each other directly—no duplication, no nonsense. The result? Faster experimentation, cleaner analysis, and a team that spends less time wrestling tech and more time executing extraordinary experiences.

And that packaged CDP? Well, let’s not beat ourselves up about it. It was the best idea we had at the time.

If you ever built your own marketing stack in the early 2010s, you might be getting a strange sense of déjà vu. Composable CDPs aren’t introducing a radical new idea—they’re just resurrecting a better one, with better tools and fewer facepalms.

1. The Data Warehouse Is the New Centre of Gravity
Back in the bad old days, your warehouse was often a read-only museum of metrics—great for reporting, hopeless for action. But today’s data warehouses are real-time, cheap to query, and built for interoperability. Composable CDPs embrace this shift. They don’t yank data into another system and force you to regurgitate it; they work with what you’ve got. Your warehouse becomes the hub—clean, central, and deeply flexible. Want to build a segment? It’s a SQL view. Launch a campaign? It’s activated via reverse ETL. Magic, minus the mystery.

2. Logging Still Matters—Especially Outbound
Marketing is no longer just about what people do—it’s about what you do in response. Propositions, campaigns, messages, experiments, nudges—all these outbound events must be logged, measured, and optimised. Traditional CDPs struggled here, often months of harmonisation and schema planning, customising for visibility or requiring third-party tools to patch the gaps. Composable stacks let you treat outbound logging as a first-class citizen. Everything from that A/B email test to your chatbot nudge gets logged back into the warehouse, ready for attribution, performance tuning, or just a smug dashboard screenshot.

3. From Platform to Protocol
Here’s a thought: what if we’ve been misreading the “P” in CDP all along? Instead of “Platform,” think Protocol. Composable CDPs don’t want to own your data—they want to orchestrate it. They define how data moves, who uses it, when, and why. Like HTTP or SMTP, it’s the rules of engagement—not another bloated destination. This shift reframes your whole stack. Your data stays where it belongs, and your tools become musicians in a well-practiced orchestra, not a group of hobbyist who’ve never met doing improv at an open mic night.

4. AI Agents Want the Same Setup
Now here’s where it gets spicy. The infrastructure you’re building for a composable CDP? It’s exactly what’s needed for the next big thing: agentic AI. Systems like MCP are all about access. Real-time, structured, and executable data pipes. They need to pull insights, make decisions, and trigger actions across your stack. Sound familiar? Composability has the same philosophy. Get your head around composability and you’ve set yourself up for the next big evolution agentic AI. More of that in another post.

So, you’re ready to leave the land of bloated platforms and build something lean, modular, and lightning-fast. Good. Here's how to do it—without getting lost in the weeds or picking a fight with your IT team.

Step 1: Define the Destination
Start by getting clear on what you actually want to do with your customer data. Better segmentation? Real-time personalisation? Smarter measurement? Your use cases should drive the tech—not the other way around.

Step 2: Audit What You’ve Got
Chances are you’re already drowning in tools and partial solutions. Sit down with your data/IT team and map out:

Step 3: Plug Into the Warehouse
Modern data warehouses (like Snowflake, BigQuery or Databricks) are the star of this show. Composable CDPs treat these as the source of truth—so your next move is connecting your marketing tools to that source. This is where tools like Hightouch or Census shine.

Step 4: Log Your Outbound Moves
Want to improve personalisation? You have to know what you’ve already said. Standardise the logging of outbound propositions—email sends, push messages, promo offers—into your warehouse. That way you can analyse what worked, what didn’t, and avoid shouting the same offer into the void three times a week.

Step 5: Think “Protocol,” Not “Platform”
Every time you consider a new tool, ask: does this play well with others? Does it use open standards? Can it be swapped in and out if needed? If not, you’re not composable—you’re cornered.

Watch Outs

Metrics That Matter

Let me be clear; if your packaged CDP is humming along, and your team has the chops to run it, optimise it, and extract value from it, then this article isn’t here to pry it out of your cold, competent hands. There’s wisdom in the old adage: don’t fix what isn’t broken.

But let’s not confuse “working” with “efficient,” or “functional” with “future-proof.”

The truth is, packaged CDPs can work really well—for companies with the size, budget, and skillset to maintain them. But even in those environments, they come with a real and ongoing cost:

And most importantly: every improvement requires working within the boundaries of someone else’s platform. You can make progress—but only as fast as your vendor roadmap allows.

That’s where the composable mindset starts to matter—even if you don’t rip anything out yet.

Start by auditing the overhead. How many hours per week are your team spending maintaining schema syncs, chasing down support tickets, or bending workflows to fit rigid integrations? What’s your true time-to-launch for new campaigns or data integrations? How instant or as we used to say “barbed wire in the baked beans email” ready are you? And are you confident that the tools you’ve locked into today won’t limit you tomorrow?

Then, look for ways to infuse composability into your existing stack. You don’t have to go full warehouse-native overnight. But you can:

This isn’t about jumping ship. It’s about future-proofing your architecture, protecting your team’s flexibility, and giving yourself options when priorities shift (and they always do).

Because just like that Ferrari—just because it drives, doesn’t mean it’s the right approach.

We started with a Ferrari stuck in first gear and somehow ended up in a DeLorean—because that’s what this shift really is. It’s not about throwing out everything you know. It’s about going back to the future we planned on before packaged CDPs: clarity, control, flexibility. Only now, the tech has finally caught up to the dream.

Packaged CDPs were a useful detour. They gave marketers power when IT was a bottleneck and duct-taping SQL into campaigns was a full-time job. But the landscape’s changed. Your warehouse is your brain. Your activation tools are sharper than ever. And composable CDPs? They let you move at the speed of now—without having to push the same data through five toll booths first.

More than anything, this is about a mindset shift.
It’s the difference between being locked into a platform that tells you how your stack works—and using a protocol that adapts to the way you work.

It’s also about opportunity. Because whether you’re a seasoned architect or a marketing lead who just wants things to go faster, the tools are now on your side. The composable model doesn’t demand perfection—just curiosity, a decent handle on your data, and the will to make your systems work for you (not the other way around).

So take stock. Kick the tyres. Run a proof of concept. Connect a tool or two to your warehouse and see what happens. You don’t need to start from scratch—you just need to start asking better questions.

TBH, in a Martech landscape where every platform claims to be THE answer, maybe it’s time to trade in that Ferrari and build your own DeLorean. Doc.

If you enjoyed this edition, please forward it to a friend who’s looking to level-up their data-driven experience game - they’ll love you for it (and I will too) ⏭️ 💌

PS. When you’re ready here’s how I can help you:

Troy Muir | The Ladder