A customer data platform unifies event and profile data into persistent customer profiles and activates them: audiences to ad platforms, journeys in email, personalization on site. We will elaborate on what a CDP actually does, how it differs from a warehouse and a CRM (the confusion that never dies), the landscape including the renames that trip people up, and whether you actually need one.
What a CDP actually does?
In order to describe what CDP does we have to break it down into four different steps.
Collect. Events (clicks, page views, purchases) and traits (attributes about the person) flow in from your website, app, store systems and other tools. Anywhere you can potentially have data around your customers and their interactions with your brand or service.
Resolve identity. The phone browser at lunchtime and the laptop buyer that evening become one profile, not two strangers. This is the technically hard part, and the reason CDPs exist as a category: matching identifiers across devices, channels and time so that behavior accumulates against a person rather than a device.
Maintain profiles. Those identity-resolved profiles persist. The CDP does not forget last month when it sees this week; each profile is a running record of the relationship.
Activate. Segments go to ad platforms as audiences. Behaviors trigger email and messaging journeys. Traits feed personalization on site. This is the payoff step, and the one that separates a CDP from a database with ambitions.
The differentiator versus everything else in the stack is the combination: identity-resolved profiles built for marketing activation, in tooling marketers can operate themselves.
CDP versus warehouse versus CRM
A warehouse (BigQuery, Snowflake, Redshift) unifies data for analysis. It is the analyst’s truth: every source joined, every question answerable, no opinion about what to do next. A warehouse is the technical capability of connecting your data.
A CDP unifies people for activation. It is the marketer’s tool: profiles resolved, segments built, audiences synced, by the marketing team, without filing an engineering ticket. It is a system designed to combine data from different sources so the marketing team can execute.
A CRM manages relationships and revenue records. It is sales’ home: contacts, conversations, pipeline, deals. A typical CRM example is Hubspot and how marketing managers are using it to automate communication with their customers.
The overlaps are real and growing. We do believe that at some point some of those providers will offer similar product features. Warehouses activate through reverse ETL. The composable-CDP pattern goes further and builds CDP behavior directly on the warehouse. CRMs sprout marketing features annually. But the jobs remain distinct: the warehouse answers questions, the CDP reaches people, the CRM runs deals.
And buying one to do another’s job is the classic expensive mistake. A CDP purchased to be your analytics truth will disappoint the analysts; a warehouse pressed into real-time personalization will exhaust the engineers. Name the job first, then shop.
The landscape, and the renames that trip people up
Segment, Twilio’s CDP, remains the developer-friendly reference point most people picture when they say the category’s name.
Salesforce’s offering has been renamed again: it is now Data 360, formerly Data Cloud, and before that Salesforce CDP and Genie, with the latest rename landing at Dreamforce in October 2025. If a vendor deck or a consultant still says “Data Cloud” as the current name, the research is stale. Say “Data 360 (formerly Data Cloud)” and you will be understood by everyone.
The composable-CDP pattern, Hightouch and the warehouse-native stacks, challenges packaged CDPs by building the same profile-and-activation behavior on the warehouse you already have, rather than copying your data into another vendor’s silo.
And older names persist in enterprise stacks: Exponea is now part of Bloomreach, Tealium carries on under its own name, and plenty of procurement documents still reference products two renames behind.
The practical rule: check the current name and owner before any procurement conversation. In this category, a shortlist ages like fruit.
Do you actually need one?
You likely benefit from a CDP when identity resolution is a real problem: customers hopping devices and channels, anonymous-to-known journeys that matter, no reliable login gluing it together. When activation needs are real-time-ish, cart abandonment triggers, session personalization. And when marketer self-service matters: campaigns move at the speed of the marketing team rather than the engineering backlog. For a CDP to make sense (financially) you need to have enough scale in your ads.
You likely do not need one when a clean warehouse plus reverse ETL already covers your activation, which is the composable route and increasingly the default for warehouse-first teams. When your identity problem is small, B2B products where everyone logs in have identity solved by the login. Or when the real pain is pipeline discipline wearing a platform-shaped costume: scattered data and unreliable numbers are not fixed by adding another platform on top of them.
CDPs and marketing measurement
Let us be precise about what CDPs are not: measurement. A CDP is activation infrastructure. In some cases, CDP is being used as just data infrastructure. It moves audiences and triggers journeys; it does not prove incrementality, and it cannot tell you whether the campaign it activated actually caused anything. Asking your CDP which channel deserves budget is asking the postman to grade the letters.
What CDPs can do for measurement is feed it. Their event streams are clean, identity-resolved, first-party journeys, exactly the raw material attribution needs, and their well-kept outcome data makes model training less of an archaeology project. That is where CDPs meet the stack MarSci builds: automated data integrations carrying marketing data into marketing mix modeling and causal attribution.
Activation decides who to reach. Measurement decides what it was worth. Run both, and let neither grade its own work. To sort the two cleanly in your stack, book a call with MarSci.
Frequently Asked Questions
A system that unifies event and profile data into identity-resolved customer profiles and activates them: segments to ad platforms as audiences, behavioral triggers to email and messaging journeys, traits to on-site personalization. The defining combination is persistent unified profiles plus activation, in tooling marketers can operate without engineering support.
A warehouse unifies data for analysis; a CDP unifies people for activation. The warehouse is the analyst’s truth, the CDP is the marketer’s tool. The composable pattern blurs the line by building CDP behavior on the warehouse via reverse ETL, but the two jobs, answering questions and reaching people, stay distinct.
Data 360, renamed from Data Cloud at Dreamforce in October 2025. Earlier names include Salesforce CDP and Genie, so the product has carried four names in a handful of years. Using an old name in a procurement conversation is the classic tell of stale research, so verify current naming before you shortlist.
If cross-device identity resolution, real-time activation and marketer self-service are genuine problems for you, quite possibly. If a clean warehouse plus reverse ETL covers your activation, or logins already solve your identity problem, the composable route is usually cheaper. Many CDP purchases are warehouse problems that went shopping, so diagnose first.
