There’s a conversation happening inside a lot of marketing and data teams right now. It usually sounds something like this:
Marketers say: We can’t move fast enough. We’re waiting on segments for weeks. Personalization is limited. We can’t test anything.
Data teams say: We’ve built the models. The data is there. Marketers don’t know how to use what we’ve already given them.
Both sides are right. And neither is actually wrong. The actual problem just hasn’t been clearly stated: more often than not, there’s no clean path between where customer data lives and where marketing teams actually work.
That gap shows up differently at every company. But the solution people reach for is usually the same: a new platform. A CDP. A data tool. Something that promises to finally connect the two sides.
Sometimes that’s the right call. Often it isn’t. What almost always matters more is diagnosing what’s actually broken before you go looking for something to fix it.
That diagnosis comes down to four questions:
- Can your marketers use customer data without filing a ticket?
- Is identity resolution an active problem or a theoretical one?
- Do your CRM and paid media teams need to activate the same audiences?
- Do you have a data foundation at all?
Answer these honestly and the tooling decision mostly makes itself. Here’s how to work through each one.
Question 1: Can your marketers use customer data without filing a ticket?
If the answer is no, and your data foundation is otherwise solid, you have an access problem, not a data problem. A new data layer won’t fix it.
Most brands come to us mid-evaluation, and what we often find is that the foundation is already there. Clean, current customer profiles exist within the data warehouse, including purchase history, churn risk, and lifetime value.
What’s missing is that marketers can’t get to any of it without asking someone else to pull it for them. Adding a new platform in between doesn’t fix that. It adds a new system to integrate, maintain, and manage, and relocates the bottleneck rather than removing it.
What teams in this situation actually need is a direct line between the data platform they already have and the engagement platform their marketing team already works in. Shorter path. Fewer handoffs. No third system to keep in sync. We’ve made the full architectural case for this in Why Most Brands Don’t Need a Standalone CDP.
Question 2: Is identity resolution an active problem or a theoretical one?
If identity issues are showing up in live campaigns today, purpose-built identity tooling may be justified. If it’s a “someday” concern, it’s not a reason to buy.
Most engagement platforms handle the basics natively. An anonymous user identifies themselves, they get matched to a known profile. That covers the majority of B2C use cases reasonably well.
Where it breaks down is when the identity problem is genuinely complex: large anonymous user populations, heavy cross-device journeys, users you’re trying to stitch together based on behavioral signals rather than a clear identifier like an email or phone number.
The test is whether you’re actively dealing with the symptoms: mismatched messages, segments that look clean in the platform but don’t reflect reality, families or households getting collapsed into single profiles. If those are live problems in your environment, that’s a real case for purpose-built tooling.
If they’re not, be careful. Identity complexity is one of the most common reasons brands end up evaluating CDPs, and one of the most commonly overestimated problems at the time of purchase.
Question 3: Do your CRM and paid media teams need to activate the same audiences?
If coordinated activation across owned and paid channels is a real operational priority that’s actively being invested in — not a someday nice-to-have — some tools support this collaboration better than others. If your paid team runs independently, this isn’t your reason to buy.
If your CRM team and paid media teams need to work off the same audience data — running email, push, and paid social in coordination — that MAY be a legitimate use case. However, before investing in a new platform, we recommend piloting features that are likely already in your martech stack, like Braze Audience Sync, or a similar functionality within your customer engagement platform.
Not only does this make collaboration easier in the platforms your team is already using, but it also helps prove out the value of stronger collaboration so you can identify whether there’s a real need for a broader investment in your tech stack to support more sophisticated use cases. If you’re seeing value in stronger coordination, a CDP may make sense for you to work off shared data across multiple campaigns.
Question 4: Do you have a data foundation at all?
If there’s no warehouse, no customer model, and no identity management anywhere, you’re in a different situation entirely.
This is the one scenario where the calculus changes. You’re not looking for a way to activate data you already have. You’re looking for somewhere to start.
In that case, a platform that combines data infrastructure and a marketer-accessible interface in one place can make sense. Just be honest about what that means: you’re building a foundation, not skipping one. As your data maturity evolves, revisit whether that architecture is still the right fit, or whether a more direct connection between your data platform and engagement platform makes more sense.
How the CDP market has changed: Agentic CDPs and Databricks CustomerLake
Before acting on your diagnosis, it’s worth understanding how the market has shifted in the last year, because the options are different than they were when most CDP evaluation frameworks were written.
The traditional CDP category — a standalone platform that sits between the warehouse and the engagement tool — still exists and still makes sense for the specific cases above. But the category is getting reshaped from both directions.
Engagement platforms like Braze are expanding what they can do on the data side — native data ingestion, event tracking, and audience management capabilities that reduce the need for a separate platform in many common scenarios.
And on the data platform side, Databricks has entered the category directly with CustomerLake — what they call an Agentic CDP. The distinction from a traditional CDP is architectural: the CDP functions — identity resolution, audience building, activation, campaign management — are embedded natively in the data platform rather than bolted on top of it. Segmentation runs against live data, so there’s no copy to sync and no drift between systems. Governance travels with the data instead of being managed separately in each tool. And the work that traditionally sits in a data team’s ticket queue — audience building, enrichment, personalization decisioning — is handled by purpose-built agents operating directly on the data, with LLM-based decisioning that makes personalization something the system can run autonomously rather than something a team executes segment by segment.
CustomerLake also represents a broader shift in the martech landscape: data platforms are becoming activation platforms. For brands already running on Databricks, it’s worth understanding before committing to a standalone CDP that may duplicate capability already sitting in your stack.
None of this means the answer is obvious. It means the decision has more options than it did two years ago, and the right answer increasingly depends on what you’re already running and what your diagnosis actually surfaced.
The cost of buying before you diagnose
If a standalone CDP does turn out to be the right call, go in clear on total cost of ownership. The license is only the visible part. Integration work, ongoing maintenance, and the overhead of staffing a third platform compound over a typical three-year contract, and they should be weighed against the cost of simply connecting the platforms you probably already have. We’ve broken down the full cost reality and the architecture behind it in Why Most Brands Don’t Need a Standalone CDP.
What your answers mean: matching the diagnosis to the fix
If you answered “no” to Question 1 and your foundation is solid, you have an access problem. The fix is a shorter path between where data lives and where marketing works, not a new platform sitting between them.
If you answered “yes, actively” to Question 2 or 3, there may be a real case for purpose-built tooling, with a standalone CDP among the options. The specifics of your environment will tell you more than any framework will.
If you answered “no” to Question 4, you’re building a foundation, not choosing an activation path. Pick infrastructure you won’t outgrow.
And if you’re on Databricks, run every one of these answers against what CustomerLake already gives you before adding anything standalone.
The question worth starting with was never “should we get a CDP?” It’s: what are we actually trying to do, and does our current stack let us do it without massive lift?
Shameless plug: If you’re working through whether to invest in a CDP, we offer a complementary martech assessment to identify possible redundancies in your tech stack and give you a clear roadmap before you sign on the dotted line. Contact us to learn more if this would be helpful.
