Don’t let your CDP become the Fall Guy

The CDP market is taking a breather.

I was recently speaking to an industry leader who told me:

“I think the whole CDP space has quietly ground to a halt… We’re seeing no more than 3–5 new opportunities per quarter. Lowest number ever.”

Funding has dried up, unless you are adding the “AI” label to a new feature. Sales cycles have slowed. Implementations are dragging. And the next big thing? It’s not here yet. AI might be “le mot du jour” (French for “word of the day”) of the moment, but it’s not the thing replacing your CDP. At best, it’s a feature set waiting on your stack to get its act together.

So what now?

Here’s what I think. I think that this is exactly the right moment to get more value out of the investment you already made. Especially when it comes to data quality, the part everybody tries to avoid like the plague, until everything breaks.

Shelfware is a data problem, not a platform problem

If your CDP feels underwhelming, vibes of which I seem to be picking up, it’s probably not because you bought the wrong tool. It’s because you haven’t given it the fuel it needs.

The hard truth is we must all face is that most CDPs don’t fail because of bad tech. In most cases, they fail because the data flowing into them is incomplete, unstructured, duplicated, misnamed, undocumented, or just stale.

And AI? Repeat after me → AI.DOES.NOT.FIX.BAD.DATA. (should .com this? 😉). It just automates more decisions based on that bad data, faster, at scale, and with even less explainability.

At some point, every CDP becomes the fall guy. (No offence to Ryan Gosling.)

But maybe we should stop blaming the platform and start fixing what feeds it.

That’s why now, in this moment where CDP development has stalled, is the time to make your data quality someone’s job again.

Five things you can do while the industry stalls

Trust me on this, you don’t need a 12-month roadmap. You need a few weeks of focused effort and some elbow grease. These aren’t sexy projects by any means. But they’re the difference between a CDP that drives value and one that quietly rots on the shelf… and gets pushed off of it.

1. Run a “data purpose audit”

Ask a simple question: why are we collecting this data point? If you don’t know what it powers, who needs it, kill it or quarantine it. You’ll reduce noise and improve downstream trust.

2. Fix your identity graph hygiene

Attention is increasing on improving Identity Resolution. Deterministic was the way to go, now probabilistic means have entered the race. But there is no immediate rush to join that conga line. Just remember, if you’re still relying on default merge rules or have user profiles with five IDs, it’s time to get serious. Resolve duplicates, revisit stitching logic, and make your identity model predictable.

3. Clean up your taxonomy

Data catalogue, anyone? Yeah, just about as well-known as some washed-up actor in a C-list movie. No offense intended, but ask yourself this → are your event names, properties, and trait labels consistent across tools? Probably not. Settle on a naming convention that survives handovers and doesn’t require an archaeologist to interpret.

4. Automate freshness checks

Which sources haven’t sent data in 30 days? Which sources do you have (my favorite question of all time to make people realize what is at stake)? Which traits haven’t been updated since last year? Set up monitoring and alerting, or accept that you’re flying blind.

5. Write it down, like someone new is joining

As a good friend of mine often says, “document or die”… well, he doesn’t say it, but he has a cool decal of it. Most CDP setups live inside one person’s head. If you are lucky, it is someone internal and not a contractor like myself. Not documenting your CDP setup is a monumental a risk. Take a moment to consider the follow → If you had to hand this over tomorrow, could someone else keep it running? If not, fix that.

It’s the foundation

Better data makes every part of your stack smarter. It fuels better segments, more reliable experiences, cleaner handoffs to downstream systems, and yes, makes AI features actually usable.

The irony? Most companies already have the tools. They just don’t have the habit of maintenance.

That’s the opportunity right now. Not a change of vendor. Not a platform rip-and-replace. But a return to basics, while everyone else waits for the next thing to arrive.

You’re already sitting on a goldmine

As I said to that industry leader I mentioned at the beginning:

“Many companies still need to realize the gold mines they are sitting on. Maybe that’s the ‘next best action’, take action while the industry is taking a breather.”

Look, your CDP isn’t broken. It’s just buried under too many things you stopped looking at.

If you’re trying to figure out where to dig or how to stop the data rot before it spreads, this is precisely the kind of thing I help clients with at Martech Therapy. One part strategy, one part therapy, and a lot of time spent with logs, identities, and unloved schemas.

Sometimes, making sense of a CDP means stepping outside the platform for a moment.

And sometimes, it means not blaming the fall guy.