The Death of the CDP?

Written By Paco Albie

I had an interesting conversation recently around the so called “death of the CDP” and how AI might replace it.

That got me thinking.

Many clients have thousands of use cases where we’d expect a CDP to generate value and while AI is advancing rapidly in pace and scale, the need for foundational tools hasn’t been made obsolete. If anything, AI exposes weaknesses in the underlying data layer.

So I decided to strip things back to the essentials:

Why would a business invest in a CDP?

To get a Single View of Customer (SVOC)

-> To deliver the right message, at the right time, in the right place to the right person

That’s personalisation.

Personalisation requires four capabilities:

1. Manipulate the presentation layer (generally in real time)

Can I access a real-time audience API and dynamically update the CMS, email platform, or media channels?

2. Know the customer (SVOC and analytics)

Can I stitch together data from multiple channels, interactions, and events into a single profile?

3. Predict what to serve (ML)

Can I apply machine learning on SVOC data to predict what the customer is likely to engage with next?

4. Have the content ready to serve (AI)

Can I generate content dynamically based on prediction outputs?

All of this is underpinned by your customer data and that’s where the CDP comes in.

To deliver personalised experiences at scale, you need the core components of a CDP:

  • Data streaming and ingestion

  • Identity resolution

  • Scalable, queryable storage

  • Data aggregation

  • Reverse ETL (activation into marketing systems)

Two end Businesses

The customer data enabled business …. able to deliver the right message, at the right time, in the right place - based on customer signals, not guesswork.

Right message tailored creative/content
Right time moments when the customer is most likely to act
Right place the channel where the customer is active and engaged

The plug it all into AI business …. without control over their first-party data are using AI to generate random content based on LLMs and web-scale training sets.

This can:

  • Fail to drive bottom-line value

  • Create brand inconsistency

  • Introduce legal and reputational risks

Paco Albie
Next

What is Identity Resolution.