A CDP (Customer Data Platform) is the software that puts everything you know about each customer into a single profile. Your customer is one person, but your systems see five strangers: an email address in the mailing platform, a cart in the ecommerce store, a ticket in support, a phone number in WhatsApp and a row in the sales team’s spreadsheet. The CDP resolves that they are the same person, keeps that profile updated in real time and makes it ready to use.
What CDP means
CDP stands for Customer Data Platform. The CDP Institute, the organization that coined the category in 2016, defines it as “software that creates and maintains a persistent, unified customer record that is accessible to other systems”. Every part of that definition matters:
- Unified: one profile per person, not one per tool.
- Persistent: history is kept, not overwritten with each update.
- Accessible to other systems: the data is not locked away; email, WhatsApp, advertising and reporting all use it.
How a CDP works, step by step
A CDP does five things, in this order. The first three organize the data; the last two turn it into revenue.
1. It collects events from every source
Every interaction arrives as an event with a date and context: viewed a page, added to cart, purchased, opened an email, wrote on WhatsApp, requested a refund. Typical sources are the website and app, the ecommerce platform (Shopify, Tiendanube, VTEX), the ERP or in-store POS, the CRM and support tools.
2. It resolves identity
This is the step nobody plans and everybody needs. The person who browsed anonymously on Monday, left an email on Tuesday and bought over WhatsApp on Friday is the same human, even though each system knows them by a different identifier. The CDP merges those identifiers (cookie, email, phone, customer ID) into one profile.
3. It builds the unified profile
The profile is not a snapshot: it is the customer’s full movie. Attributes (country, plan, signup date) plus the entire behavioral history. That is where the metrics that matter are calculated, such as purchase frequency, average order value or lifecycle stage.
4. It segments in real time
Segments update themselves. “Bought twice and has not returned in 60 days” is not a list someone exports on Mondays: it is a condition the CDP evaluates continuously, and a customer enters or leaves the segment as soon as their behavior changes.
5. It activates channels
Flows run on top of those segments: the abandoned cart, staged onboarding, dormant-customer reactivation, the audience synced to Meta Ads. This is what separates a CDP from a data warehouse: the warehouse stores data and answers queries; the CDP also acts.
CDP vs CRM vs DMP vs data warehouse
The question we hear most is “I already have a CRM, why would I want a CDP?”. The four tools get confused because they all store customer data, but they solve different problems:
| CDP | CRM | DMP | Data warehouse | |
|---|---|---|---|---|
| Data | Behavior of identified and anonymous customers | Commercial relationship: contacts, deals, notes | Anonymous audiences, cookies, third-party data | Any business data |
| How it is fed | Automatically, by events | Manual entry by the team | Pixels and data vendors | Data engineering pipelines |
| What it is for | Segmenting and activating communication | Managing the sale and the account | Buying advertising | Analysis and reporting |
| Who runs it | Marketing and growth | Sales and service | Media teams | Data team |
| Keeps history | Yes, persistent | Partially | No, it expires | Yes |
The costliest difference to ignore is the CRM one. A CRM depends on human input: if nobody enters data, there is no data. A CDP feeds itself: every click, purchase and message arrives without anyone typing. A B2B business with a long sales cycle lives in its CRM. A business with thousands of customers who buy, return and churn needs a CDP, because no team can manually record the behavior of 100,000 people. And they work together: the CRM usually connects as one more source of the profile.
The DMP confusion is generational. DMPs relied on third-party cookies to buy advertising; as those are phased out, the first-party data a CDP accumulates has become the central asset.
Types of CDP
Not every product sold as a CDP does the same job. There are three families:
Packaged CDP with activation
Collects, unifies and also runs campaigns in the same place, as Customer.io or Bloomreach do. It fits when the marketing team has to operate without depending on a data team: the same profile that gets built is the one that triggers the email or the WhatsApp message.
Event collection and routing CDP
Captures events, organizes them and sends them to other tools. Segment and RudderStack are the classic examples. They are the plumbing: very flexible, but activation happens in another tool.
Composable CDP on top of the data warehouse
Uses your warehouse (BigQuery, Snowflake) as the base and syncs audiences to channels, as Hightouch does. It makes sense when there is already a data team and the warehouse is the source of truth.
For mid-size companies in Latin America and the US Hispanic market, the combination we see work best is the first one, sometimes with a warehouse alongside for analysis. If you are comparing options for an online store, see our Klaviyo vs Customer.io comparison for Shopify.
Use cases by industry
A CDP is not only an ecommerce tool: it fits any business with many customers and a repeating cycle.
- Ecommerce and retail: abandoned cart and browse, repurchase timed to each product’s consumption cycle, dormant-customer reactivation, and joining online and in-store sales.
- Fintech and apps: activating the user who signed up but never made a first deposit, staged onboarding, churn prevention based on inactivity.
- Automotive and assisted sales: following leads that arrive over WhatsApp through to the sale and after-sales service. At Grupo Garden the CDP became the single source of truth for 10 car brands.
- Subscriptions: failed payments, renewal notices, win-back of cancellations.
Signs you already need one
- Every platform reports a different sales number and meetings are spent arguing about which one is real.
- Your campaigns send the same thing to the customer who bought yesterday and to the one who has not returned since 2023.
- The personalization you want requires “asking IT for the data” every time, and it takes weeks.
- WhatsApp is your most effective channel but runs disconnected from everything else: nobody knows what conversation each customer had.
How much a CDP costs
The cost has two parts, and the second one is the one people underestimate:
- The license. Charged by profile, event or message volume depending on the vendor. Several publish their pricing; others quote case by case.
- Implementation and operation. Connecting sources, defining events, resolving identity and building the flows. A good implementation takes weeks, but operating the CDP is a permanent weekly job. That is why so many companies buy a CDP and end up using it as a mailing list.
How to implement it without a never-ending project
The classic fear is the endless project. Reality depends on the method: with event schemas and a proven flow library, the Bigbox implementation reached production in 15 days and was then replicated across five countries. The order that works:
- Define the event schema with the business’s names, not the software’s.
- Connect sources in order of value: ecommerce first, then email, support and offline.
- Resolve identity across systems.
- Migrate existing flows first, then build new ones.
- Measure against a control group from the very first flow.
How to choose a CDP
Before watching demos, answer these five questions:
- Who will run it? Without a data team, you need a CDP with built-in activation.
- Which channels do you need to activate? If WhatsApp is central, check that the integration is native or via API, not a workaround.
- Where does your data live today? If there is already a well-kept warehouse, a composable approach may make sense.
- How does pricing scale? By profiles, events or messages: project your volume two years out.
- Who will implement it? The tool is half the result.
At Hyppo we are a data marketing agency and Customer.io partners: we implement the CDP and operate it every week, with growth automation as the complete system. The platform is half the work; the daily operation on top of it is the other half, and it is where a CDP goes from installed software to a growth machine.