In search for ideal CDP for B2B SaaS
And do you really need one.
If you were forwarded this newsletter, consider joining over 200 weekly readers—some of the smartest GTM devs—by subscribing here:
Imagine this scenario: You're building a SaaS product. Your engineering team is strong, customers adore your product, and your sales team is closing impressive deals. Then Emily, fresh from her role at a %famous and successful SaaS startup%, joins your marketing team and asks, "What CDP are we using?"
And then you're like – “What exactly is a CDP”?
Historically, CDPs (customer data platforms) have been the realm of marketing folks at large B2C orgs, primarily used for customer segmentation and identity resolution to refine ad targeting and enhance online engagement.
When I first started this post and took a closer look at the CDP vendor landscape, I was surprised by just how fragmented it is. There are CDPs for telco built by telco, CDP for fans engagement , CDPs for the CN market, CDP for NL market, CDP for website on Drupal, CDP for fashion, CDP for restaurants and etc.
Marketing operations are unique in every business, differing in tech stacks, processes, and regulations, creating opportunities to tailor tools for specific niches.
Twilio CEO Jeff Lawson explained to Stratechery’s Ben Thompson after acquiring Segment that the abundance of SaaS solutions has led to data being trapped in silos across different business lines. Marketing data resides in multiple platforms (social media, email, CRM, advertising tools), making it difficult to integrate.
A premise CDP is simple – it aligns your operational stack, enabling it to function as a cohesive unit by consolidating both data and crucial workflows.
In the B2B SaaS the workflows typically fall within the realms of sales, marketing, and data teams. Moreover, the complexity and the wide range of opportunities of CDPs often overlap with other three-letter acronym systems like CRM (customer relationship management), CEP (customer engagement platform), ABM (account-based marketing), or MAP (marketing automation platform).
I teamed up with Aurelien from Cargo to help with this post. With his team, he is building the future of the gtm stack, empowering revenue teams to create AI-powered workflows that automate everything from account identification and enrichment to scoring and routing.
Packaged approach
Originally packaged CDPs were designed primarily focusing on event-clickstream data and help to map customers across distinct systems. Normally the tools come with built-in segmentation tools, making it easy to create customer segments and immediately activate them in marketing channels, customer service platforms, or analytics dashboard
Packaged CDPs often feature built-in compatibility and integrations with popular downstream tools, that close any of the following competences:
Customer Engagement: email, push notifications, in-app messaging, and SMS campaigns
Product and Behavioral Analytics: data federation, event tracking, schema management, and data privacy compliance
Attribution: tracking for paid and organic channels, campaign performance analysis, and multi-touch attribution models
Customer Acquisition and Growth Marketing: SEO, SEM, affiliate marketing, referral programs, and deep-linking for customer acquisition
Data Infrastructure and Routing: event streaming, real-time routing, and API integrations for seamless data flow
In 2024, the essence of packaged CDPs remains consistent across various vendors. The fundamental premise is straightforward: we integrate all your data sources, enabling you to orchestrate data across various services—albeit limited to those for which connections you have. The real differentiation among these packaged CDPs lies not in their basic functionalities, but in what happens within this 'black box' of data processing, for example, some fancy ML-powered segmentation. Also, it’s about the degree of control you have over data federation – specifically, the customization of triggers, workflows, and other operational elements.
The black-box identity resolution is especially crucial, as vendors typically use some combination of probabilistic and deterministic methods for identity resolution. However, these approaches may not be efficient when handling complexity of entity models in various B2B SaaS environments, especially in collaborative and group scenarios. Especially in PLS products you design you ledder in a way to scale from individual user to company usage.
Originally CDPs were never intended to serve as the single source of truth for data. But the latest trend in packaged B2B SaaS CDPs is the blurring of upstream and downstream roles. Platforms historically seen as engagement tools are now evolving into dominant CDP players. More tools than ever now offer core capabilities like data collection, federation, and audience creation—like Amplitude, Rudderstack, Snowplow, and Braze. Integrations and building on top of data warehouses to keep data where it resides have also become the norm.
Composable approach
Data Warehouses become more common among B2B companies keen on exploring the so-called modern data stack. Initially, these efforts predominantly focused on BI and analytical needs. However it often resulted in the data being used in relatively static ways.
Tools like Hightouch and Census emerged as “reverse ETL” tools—sending data from the data warehouse back to SaaS tools (Salesforce, Marketo, Facebook Ads, etc.) without coding the infrastructure for that.
The technology behind these tools may not seem highly sophisticated at first glance, but they excel by developing and maintaining schemas for hundreds of connectors and data sources. Reverse-ETL emerged as the missing piece that made bridge the data and GTM world, capable of serving as the source of truth for all customer engagement data. Since then, both Hightouch and Census evolved into a suite of tools around the warehouse (identity resolution, data enrichment, event streaming, etc.).
MarkTech community soon realized that they could effectively "de-package" the packaged CDP by using separate, best-in-class tools:
ETL for data collection and transformation
Data Warehouse for storage and centralization
Reverse-ETL for data distribution and activation
*name the category* for identity resolution, data enrichment and etc
The Composable CDP approach provides a solution to the perennial 'Build vs Buy' dilemma, suggesting a 'Build and Buy what you don't wish to maintain' strategy.
Composable CDPs differ from Packaged CDPs in their customizable nature. While packaged solutions are pre-built with certain integration and adaptability limits, composable CDPs enable businesses to tailor an architecture to their unique needs without acting as a 'black box' for data storage and management.
This flexibility is invaluable for B2B firms tracking distinct sales funnel stages or SaaS companies integrating product usage data into CRM systems.
Key capabilities of Composable CDPs include:
Deterministic identity stitching and probabilistic identity resolution.
SQL-based data transformation pipelines.
Configurable data cleansing.
Less vendor-lock
This approach enables the formation of a single entity across various operational systems. Consolidation is key to performing focused analyses on specific entities like customer churn or annual recurring revenue, using unified semantic models across all SaaS tools.
While composable CDPs may seem like developer-only tools, they’re actually lowering the barrier, making these platforms more accessible to non-technical users.
Additionally, the lines between packaged and composable solutions are becoming increasingly blurred. Packaged solutions are focusing on tools for data users, while composable options are evolving in a different direction. This evolution aims to combine the best features of both, resulting in what some are now calling a hybrid approach.
CDP in B2B SaaS
The funny thing is, I started writing this post in 2023, and now, revisiting it, I’ve had to radically cut much of what I originally wrote. And so for the Q4 of 2024, I feel that In the B2B SaaS context, a CDP in its original form might not be essential at all. While there are some dedicated B2B CDPs, they often function as "bolt-ons" to existing ABM or MAP platforms. This means they don’t really function as an independent layer in your marktech stack; instead, they serve as extensions or add-ons to current offerings.
In the SaaS landscape, many tools blur the lines of what a traditional CDP provides. Honestly, I don't think there’s much room for a packaged approach here. If you’re a growing SaaS you either use CRM along with marketing automation systems, or you go all-in with a composable approach that allows you to tailor everything to your specific data model needs.
In B2B SaaS, the long-discussed intersection of sales, product, and marketing teams has become a reality. Marketing and sales no longer operate in silos. We’re witnessing a redesign of ABM platforms, opening the door for a new generation of tools that replicate the original role of the CDP.
This new approach is driven by smart enrichment, outbound automation, workflow orchestration, segmentation, and data modeling tailored to B2B nuances. It may even lead to an entirely new category—GTM tech—to meet these evolving needs.
However, I believe at least for now the original premise of CDPs in SaaS will rest on the shoulders of a new breed of professionals—RevOps and GTM engineers—who combine technical expertise with business acumen.