Why composable CDPs benefit CRM teams

As I wrote in an earlier article, the benefits of composability have often been expressed in terms of the architectural principles of data management and how this benefits data teams.

While these benefits are important, I’d argue they’re not the primary benefits of composability.

Businesses ultimately invest in CDPs to drive marketing and customer experience performance metrics – improving customer acquisition efficiency, increasing customer lifetime value, driving up purchase frequency, etc.

Expressing the benefits of composability purely in architectural terms has limited conversation around how it benefits the above business outcomes expected of a CDP and the customer marketing teams that invest in the process.

Martin Kihn at Salesforce, the way less cool version of Don Cheadle, recently published Salesforce’s future vision for their Data Cloud, a term they now apparently use interchangeably with CDP.

While I won’t go into their CDP approach in this piece (and I mostly expressed my opinions on where Salesforce’s Data Cloud is headed in my previous article), Salesforce’s vision for the capabilities a CDP should offer (if you look past the vaporware and marketing jargon) underscores a few key business benefits of composability, even though this is clearly not the approach Salesforce is taking with their Data Cloud.

First, let’s talk about the composable trend and why it’s important.

Hightouch started creating noise around this term in 2022 with aggressive claims that the CDP is dead and this piece explaining the benefits of a composable CDP vs. a traditional CDP.

So, what exactly is Hightouch’s stance?

Hightouch was founded by a group of former Segment engineers who attached Hightouch’s vision to data warehouse centricity and the #moderndatastack. It’s also worth mentioning that Segment seemed to miss this data-warehouse-centric trend and now Twilio’s business is feeling the pressure.

I agree with just about everything Hightouch has to say about traditional CDPs. Traditional CDPs predominately come from an era of SaaS where the application wants to become the source of truth for whatever data it controls.

Salesforce pioneered this approach by creating a cloud for everything that stores an authoritative copy of your sales, marketing, advertising, support, and [insert data for any use case here] data. Data Cloud is now their response to the problems this approach creates.

This approach means that the data marketing teams are working with is often siloed from investments being made on the data side for other use cases.

A great example of this is that enterprises will often have analytics and data science teams interpreting data to gather insights and that same work is rarely being used or seems to seldom be available to support marketing use cases.

At Simon, we’ve always embraced data warehouse centricity and the myriad benefits centralizing data within a data warehouse creates for a business, beyond CDP use cases.

Hightouch is a product built by engineers for engineers (like Segment) and so the vision and product marketing around it have always focused on the benefits for engineers. My goal here is to translate some of these benefits into the things that matter most for the teams that use data to power the customer experience.

If the Cloud Data Warehouse is the center of your data strategy, and the CDP is fully connected to your CDW, there are many benefits for customer marketing teams beyond the benefits for data teams that Hightouch and others have expressed.

1. Auto-optimizing experiences

If the CDP is fully connected to all customer data — think ML models trained on historical data, engagement data, real-time data from web/app, etc., — the CDP can auto-optimize the customer experience.

Why is this different in a composable approach? The answer is simple: the optimization is driven by the highest quality dataset (i.e., the complete customer profile).

When real-time and historical data come from different sources and the CDP is organizing this (a classic traditional CDP use case), ML models may take into account something like the customer’s historical lifetime value, but it cannot fully incorporate real-time data in training the models that predict what experience to serve the customer.

2. Data availability

One of the things I hear most often from CRM teams is that they spend a significant amount of time wrangling and structuring data to support their various marketing use cases.

This is primarily driven by the rigid data structures of marketing automation platforms (I said I wouldn’t continue to talk about Salesforce in this article, but SFMC data extensions or PET tables in Oracle Marketing Cloud are probably the best examples of this).

With a composable approach and well-structured data in the data warehouse, the CDP can function as the “brain” of marketing automation, with the downstream platform simply executing the messaging.

3. Identity management

Customer identity is the biggest blocker to achieving fast time to value with a CDP. For enterprises with complex data structures (e.g., accounts, households, multiple brands, etc.), this is especially true.

While engineering-focused rETL tools talk about this benefit, and composability provides significant benefits in enabling teams to leverage complex identity structures, these benefits should also be extended to CRM teams.

Getting identity right means not only being able to structure data to reflect the customer context but also being able to personalize the customer experience against it (and leverage relational data tied to the customer context).

For example, we have many clients who have multiple brands, and being able to personalize the experience based on the context of a given brand, while also being able to understand the customer holistically across brands, is an important use case that comes to mind.

With a traditional CDP, and Salesforce Data Cloud may now be the most prominent example of this, the quality of your identity resolution, data science model outputs, and the blocking and tackling of your CRM program — segmentation, personalization, orchestration, and the like are dependent on the data quality within your traditional CDP.

Why composability matters

With a composable approach, there is no divide between investments designed to solve these problems at an enterprise data level and the business outcomes that CRM teams are driving toward.

Further, composability means that achieving these goals can be accomplished through credentialization of the CDP into the data warehouse vs. a months- or years-long implementation of the CDP and ongoing maintenance and integration work as data changes.

With GenAI advancements, GPTs and the like can interpret a dataset to, for example, generate the right email content for 1-1 personalization. GenAI could do this regardless of where the data lives, but the quality of the output is going to be determined almost entirely by the quality of the dataset and not the quality of the model itself — something that is certainly improved with a composable approach because the GPT can interpret the entire corpus of customer data.

I predict that the more vendors, data, and CRM teams adopt composable approaches and the more they can express the business value or use cases that composable approaches enable, this trend will move from the collective industry navel-gazing that composability seems to have so far represented and will eventually become a requirement of CDP vendors.