CDP (Customer data platform)

What a CDP does

Let's take a shoe workshop in Mexico as an example. They make leather shoes and sell them online to Mexico and the United States. A customer can buy on the site, sign up for restock emails, click a Meta ad, and write to support about a size. Those events often live in four tools with four IDs. A CDP's (customer data platform's) job is to decide they are one person, then let email, ads, and the site use that one profile.

A customer data platform does three jobs. It ingests data from many sources. It resolves identity, deciding that an email address, a logged-in ID, and a device belong to the same person. It exposes the result to other systems so audiences stay consistent. Walk a shopper at a shoe workshop. She bought oxblood loafers last March under one email. She clicked a sandal ad in June on a different device. Without a customer data platform, the workshop retargets her with the loafers she already owns. With one, the sandal ad can exclude that purchase.

The value of a customer data platform is in the second job. Ingestion is plumbing. Identity resolution is the part that turns four partial records into a customer you can actually measure LTV (customer lifetime value: what a customer is worth over time) on.

CDP vs CRM vs data warehouse

A CRM (customer relationship management system) is built around a sales relationship and is usually edited by humans. A salesperson updates a deal. A data warehouse stores everything, in any shape, for analysis. A CDP (customer data platform) sits between them. It is opinionated about the customer profile and built to push audiences out to marketing tools in near real time. Let's take a shoe workshop in Mexico as an example. They make leather shoes and sell them online to Mexico and the United States. The workshop does not need a salesperson to tick a box before a shopper drops out of the loafer audience.

Warehouse-native customer data platforms blur this by keeping the data in the warehouse and running resolution and activation on top. That is now the common pattern rather than the exception. The label on the vendor matters less than whether a shoe workshop can exclude known buyers from a prospecting ad.

What a CDP will not fix

A customer data platform will not fix attribution (which ad or channel gets credit for a sale). Knowing a shopper is one person does not tell you whether the Meta ad caused the sandal order. It will not fix bad event tracking either. A customer data platform resolves identity on the data you send it, so if the purchase event fires twice, the profiles are confidently wrong.

A customer data platform also will not make anyone act on the data. Most teams that bought one still export to a sheet on Monday morning. Let's take a shoe workshop in Mexico as an example. They make leather shoes and sell them online to Mexico and the United States. The workshop still needs a person (or an AI co-worker) to use the audience. Lemonado is not a customer data platform. It connects to the tools you already run and does that work on top, including writes in connected ad accounts when you ask. MCP (Model Context Protocol) stays read-only.

Common CDP mistakes

• Buying a customer data platform before a shoe workshop agrees on what a customer is: email, household, or logged-in shopper.
• Piping every event in without a schema and calling it a single source of truth.
• Expecting audience quality to fix a creative problem on the sandal ad.