What is a Customer Data Platform
A customer data platform (CDP) is software (homegrown or purchased) that collects and unifies all your customer and prospect data from multiple sources building a single, coherent, accessible, complete view of each customer.
I know what you are thinking… “I have something that does this already.” Well I used to be in the camp of “Why do I need a CDP? my MAP or CRM or Data warehouse pulls together all my customer data. Each of those solutions has benefits over the others, for the usages scenario it was designed to address. But the CDP has some unique features to offer.
As we add more tech to the MarTech stack, functionality seems to start to overlap and that can definitely be true for some products. This time the nuances make all the difference. We will discuss the concept of a CDP rather than any particular technology or product.
Benefits of a CDP
No data is lost
Organizes and makes your data coherent
Unifies data at a person level
Deep understanding of customers and prospects
Seamless cross-channel customer experiences
Let’s talk about some Customer Data scenarios you might have experienced.
Scenario: You are Saved… No Data is Lost
one of the marketers updates a form on a web page and renames a field rather than the label…
depending on the system, the form is hosted on… most likely the data is lost
CDPs have a solution for unmapped fields so no data is lost
Scenario: Rich understandable profile at the person level
you get a complaint about an email sent to a customer from a salesperson… all you have is a name and company.
so you start with your MAP and look up the name and company to try to get at an email address and you find a couple of email address records with the name/company combination
looks like there is an email address for their work email and general work email (info@acme.com) and 2 personal emails… not exactly straight forward but you dig into recent activity to see what was sent but you don’t find anything
maybe it was a sales email… so you dig into your CRM… you find multiple records and you look at the task and activities and you don’t find anything
Racking your brain… you come up with another place to look… Customer Success and Training… sometimes they promote training events in their systems
CDPs can solve this will a unified record at the individual level where the data is rich and understandable
Scenario: Seamless Cross-channel experiences
as a marketer, you want to have a consistent message (maybe not the exact same but a consistent theme for your prospects) across all touchpoints
but you find… email system is using email to identify users, Website is using a cookie ID in some places and a login ID in authenticated areas, social depending on the platform uses a mix of personal and work email addresses… and your training systems have student ID and often the email a personal email because the participants want to carry certification with them wherever they work
stitching this together is possible but painful… CDP can make this easy and seamless
If we are honest, we have alot of the same data in many systems and some time in multiple fields in the same system … this is especially bad in systems that have been live the longest… I’m looking at your CRM, MAP, etc Most of us don’t have a truly centralized customer database that can be easily accessed and queried to gain a deep understanding of the customer.
Key Features to look for in a CDP
Let’s start with the basic that all CDPs should have and then dive into some of the different types of CDPs
Easy and Fast Integrations
as many of you have likely experienced, custom integrations can be a pain
CDPs have lots of ready-build integrations to pull and push data to the most popular apps
for campaigning, analytics and data science, sales follow-up, Customer service, etc
obviously this varies depending on the product
These let you gather data as well as push processed data … to/from disparate tools from across your enterprise’s tech stack… not just in marketing
Campaign execution systems can also quickly pull data from CDPs for coordination, personalization
Automated Centralized Data Processing
at the core of a CDP it quickly processes the data it’s gathered, classifying, categorizing and transforming data
This often includes: Firmographic, Technographic, audience membership, persona, campaign membership, behavioral, Deal info, entitlements, and support cases to name a few
all connected together with persistent ID
when merging data you will find there are conflicting record and values from your various sources and Fast Data processing (matching and transformations) is required to make the platform viable as a central data hub
Sometimes as source systems are updated there may be new unmapped data fields or values, CDP don’t need data to be organized but it’s prefered (to keep data organized)
they can accept new data fields and values ensuring you don’t lose data so you can deal with them after collection
don’t let this run too long or you will end with a mess
Centralized Coordination
Traditionally the owner of the system that collected the data decides what are the next steps in terms of actions and data flow
As we hinted earlier, CDP centrally connect systems that don’t have native integrations with each other orchestrating how data flows and is processed
Because a CDP identifies the same person even if they’re using different emails (or other identifiers), and ties data from across the funnel to one persistent customer identity we can also orchestrate business actions (with segments, personas, setting flags and stages etc)
depending on the type of CDP you have you can also have a central place for you to pull organized clean data for analysis and Data science…
data clarity is often underappreciated until you find 5 difference fields that seem to have similar data
Types of CDPs
As we alluded to earlier not all CDPs have the same features. There are 3 types of CDPs… from the simplest to the most feature rich:
Data Connectors - pipe data from one tool to another but don’t retain it. These are easy to implement… setup the connection on each end and the logic for transformations in between (i.e. Segment and MetaRouter and many new offerings sit here)
Store and Analyze - are the standalone traditional CDPs, that store and process data as well as communicate with other tools. They are the centralized source of truth on customer data and analytics (i.e. Lytics and Tealium)
Orchestrating - do all of the above, plus real-time segmentation, individual content recommendations, campaign execution triggering, and more (i.e. BlueShift)
CDP vs CRM, MAP, MDM
All of these systems have been expanding their functionality to address the same need but from their perspective. This is where the nuances make a huge difference. Whether you choose to buy or build a CDP these are the differentiators to include.
All CDPs accept all incoming data and create new fields on the fly linked to the record
this is because most CDPs use a data structure called a non-relational database
non-relational databases means data can be added first, and sorted later
on other types of platforms, most often the unknown fields might be stored temporarily in a log or on a special table but not linked and easily accessible from the main record
With so much data going the the CDP, one of the core abilities are complex/granular data rules to determine what data is updated and how, and what is passed to other systems
for example field level changes prompt that field to be sent to other systems not the entire record
this is really important to ensure we don’t cause what I like to call data echos, this is when you have a value update in one system… but you have to push the entire record… there is an out of data value in another system and it comes back as a updated value because the entire record has a newer date and the value is different than the current value (bad)
most of other systems only timestamp changes at the record level (at least in a usable manner… as logs are often not readily accessible and usable)
Real-time data updates (especially when view with field level updates solves a number of problems)
this ensures you have the most recent data for your triggers, logic, and automation
especially important for content recommendations to know what has already been consumed
also spreads out the data transfers throughout the day and since it only send data that’s been updated less data needs to be sent in total
many of the other systems have scheduled updates and push entire records
For me this is the biggest differentiator, It identifies the same person even if they’re using different emails, and ties data from across your tech stack to one persistent customer identity
this matching and identification isn’t a one time process and as new data is added to a record and matches are found the records are in effect “merged”
the complex part of this is how you define the rules of what identifies a unique person, this is a delicate balance of including enough to find matches with various pieces of data vs too many requirements that matching doesn’t happen often enough
The other platform typically use a simplistic identifier like an email address or allows duplicates and puts the onus on the users to find and fix
