Briefly in advance:
This post will help you to understand the workflows and capabilities of Data Cloud for Marketing if you have no or just little experience with Salesforce Data Cloud or Customer Data Platforms. I will use the terms CDP (Customer Data Platform) and Data Cloud here as synonyms as I am looking at it from a marketing perspective and the features that are available today.

The power of Data Cloud

While there is a good understanding of the capabilities and value Marketing Cloud Engagement offers, there is a lot of sales and marketing blah-blah out there about CDPs, pointing out advantages nobody asked for. Nethertheless I think it is time to understand CDP not as a temporary trend anymore. Here is what makes Data Cloud so powerful.

Core & Data Cloud

Technology is the main differenciator. At Dreamforce Bret Taylor used the phrase “Marketing on Core” for this shift. It allows a more direct integration to sales, service and loyalty data and enables us to benefit from core platform capabilities like the advanced user and permission management, Lightning Experience, Flows and more. Core apps and integratios are getting feasible for marketing users.

But why didn’t Salesforce bring Marketing Cloud to Core earlier? The technical foundation and processes of Marketing Cloud Engagement are just very different to the Salesforce Core Platform. This is comprehensible when you consider the requirements we place on both solutions. Sure, also the fact that they initially have been different companies using different technologies is one of the reasons.

To make marketing possible on Core there needed to be an extension. Salesforce is using a lakehouse architecture that combines scalability and speed. In the same way as Marketing Cloud will benefit from Core capabilities, Core applications will benefit from Genie. It will have a great impact for many Salesforce products. That’s why the technical foundation of CDP became its own solution that is now Data Cloud, Salesforce’s new platform for high performant data management. CDP was just the first Salesforce product using the Genie/Data Cloud capabilities. Capabilities I will describe further in the following.

Real-time vs. Batch

When we talk about data transfer & data processing, we usually need to choose between (near) real-time processes that have fixed structures and limited data volumes and more slowly processes which on the other hand bring you flexible data models and big data management. Marketing Cloud Personalization (formerly Interaction Studio) is a good example for a system that counts real-time decisioning one of its biggest strengths but at the same time has limitations in the number of customer attributes, identifiers or related objects.

Many CDP vendors are having a similar decision to make. Some tools were developed based on web tracking or tag management solutions that often have their strength in real-time processing, capturing data in form of events and reacting in milliseconds. Other tools have their focus on storing huge amounts of data, complex data models and segmentations with a more batch-like approach. Everyone can imagine that processing millions of records across several joined objects will take more time than making a decision for only one profile record at a time with a limited complexity of related information. Vendors have to choose their sweet spot. This was no difference for Salesforce.

Salesforce CDP could be described as an Analytics CDP with the capability to reflect your very own data model and create insights and segments on it. For more real-time decisioning use cases, you most likely use Personalization (Interaction Studio) today.

For me this classification changed with the announcements of Genie and new capabilities in 2022. Instead of focussing on either data volume or speed, Salesforce aims to provide both ways in parallel. On the one hand you have streaming ingestions, Data Transforms to pick up and transform records in real-time as well as Data Actions, a functionality based on customer related events that can trigger Core Platform Events, call a webhook or send a record to Marketing Cloud. On the other hand you have batch ingestions and Data Prep Recipes to do complex transformations, join and filter objects on a scheduled basis. In addition to both, with the Spring 23 release you become able to concatenate processes. For example you can trigger Flows that update and publish a segment directly after new data has been ingested.

This maybe reminds you of Marketing Cloud, where you also have multiple ways of realizing a process (like email sends) depending on the required speed or data.

Data Model

Defining a good data model for all your operations in Data Cloud is the main topic in each implementation. The challenge is to not only copy data from one tool to another but to create a common representation of all your customer related data independently from its source or target. You dont want to rate engagement differently just because it has been ingested from one or the other source or limit decision critera to CRM data when you would have the capability to include much more information. Through mapping all your source data on common Data Model Objects you enable new segmentation options that have not been possible before.

Different to Marketing Cloud you can now involve millions and billions of events into your segmentation criteria without worrying that timeouts will kill your process. These new insights will level up your segmentation and personalization. Audiences are changing from “people who bought a product” to “people who bought a product, interacted on emails of this product, had a service case about the product, viewed webpages about the product or have colleagues who bought that product and have a product interest score > 50” in just a few clicks. Sure having more data is not only useful for extending audiences but also to exclude, focus and split them to limit spendings on irrelevant recipients or topics. In each case a good data model is crucial.

In Marketing Cloud we are used to join and copy data as much as we need. Every campaign gets its own Data Extensions, maybe even specific data deliveries. Data hygiene, reduction or uniqueness are rare topics in Marketing Cloud Engagement. In Data Cloud changes on the data model can impact existing segments and calculations which leads to the requirement to really understand your data objects and relationships.

Unification

It is a common misunderstanding that CDP would create a golden customer record that you then use to update all source and target systems. This is not the case and often confuses people as they thought this would be the main purpose of the tool. Instead CDP creates a unified view on your customer. The simple reason for that is, that just overwriting data with the “best” value often is not what you or your customer want you to do. Sometimes there are legal reasons why you are not allowed to just equalize customer data across systems or maybe your customer is having different profiles on purpose. They may expect communication about one topic on one channel and another communication on another channel. And sometimes you just cannot make 100% sure that it is the same person.

Data Cloud avoids these issues by not overwriting data but by creating mutable links between customer profiles. The unification ruleset in Data Cloud is just a different use case than what you probably would do with a MDM (Master Data Management). Because of this Data Cloud also does not create some kind of master ID that you share with other tools. It has to be independent from a universal ID or single primary key to enable its insights and options for data exploration.

This way the unified profile is giving you the capability to access customer information from each source profile and also to change the unification logic anytime if needed. You achieve the holy 360° view you want to have for segmentation.

Channel Agnosticity

Doing segmentation in Data Cloud enables you to share the result (segment) with various tools and channels. Segments are independent from the channel where you want to use it. Unlike what we know from Marketing Cloud, a Data Cloud segment does not include all the attributes that we need but is just a ruleset that concludes in a selection of individuals.

In Data Cloud Activations you choose a segment and publish it to your engagement tools, adding the specific data every tool needs. You publish contacts to Marketing Cloud Engagement using the Contact Key and email address that MC already knows. At the same time you activate the segment to an S3 bucket, to different AppExchange partners, or use it for an advertising campaign, adding different identifiers and attributes for each target. Activations share a composition of unified data (e.g. name from CRM, most frequently stated birth date, most recent specified job title) and the source identifiers (e.g. Contact Key and email address from Marketing Cloud). This ensures to have consistent identifiers in the connected tools without the need to adapt your data model for Data Cloud.

Integration & Data Access

Most Salesforce tools already have an out-of-the-box connector for Data Cloud. The best way to connect non-Salesforce products with Data Cloud is directly via API, a middleware like Mulesoft or cloud storage (Google, AWS). The Ingestion API comes with a batch and a streaming option. Not covered today is the file based way via SFTP.

After creating unified profiles and doing awesome data modeling and segmentation, you want to get the insights out of Data Cloud. The main way of doing this is what Salesforce calls “Activation”. Pushing an audience (segment) to one of your Activation Targets. Instead of activating segments, you can also query data using Query API. This is the standard way to integrate with visualization & BI tools like Tableau, MC Intelligence (Datorama) and CRM Analytics (Tableau CRM). Of couse they have native connectors to CDP. Tableau is even able to access data directly with the new Data Cloud for Tableau.

In addition, integrations have been announced to leverage data from your own data lake within Data Cloud (zero copy) and the other way around as well as the option to bring your own AI processes into Data Cloud.

Salesforce will surely offer connectors to more solutions in the future. But even today people are able to share much valuable data within their companies via the available interfaces.

Empower Marketing (Cloud)

The task for CDP/Data Cloud is a big one: Bring Marketing Cloud to Core and be technically better, more advanced and more user friendly as doing the same stuff in Marketing Cloud Engagement. At the same time Data Cloud has to replace Audience Studio, Audience Builder and possibly more MC components in the future. People ask “Is CDP just doing the same as the old tools, only with a different name?” – no. Sure it addresses some similar use cases, especially data management and segmentation. But the purpose and impact of the solution is different after all.

CDP is a System of Insight

Customer Data Platform creates value by making data applicable. This is something that is hard to measure. What we can measure is the impact of it. Here we see the differences to the approach we had for many years.

In the past we mostly had two components: a system of record, our CRM, and a system of engagement, Marketing Cloud. The system of record was responsible for providing all the data we need. The system of engagement to create audiences, manage content and personalization and deliver our messages to the customer (right message, right channel, right time and so on). Therefore Marketing Cloud comes with a set of features which of course includes segmentation. This combination worked well for many years.

The situation changed since there is not only one system of record anymore but many, storing different kinds of customer data and identifiers. A popular approach to deal with this condition is to choose one master system for customer data. But CRM as well as many other tools were never thought to be responsible to manage all your engagement data and customer identities or to act as a middleware for Marketing Cloud. The marketing data is often not used in these tools and so the tools are not the right place for it.

At the same time companies communicate via different marketing channels that require the use of spezialized tools and interfaces. Segmentations often are done in various tools based on the same criteria to deliver consistent messaging across different channels. Therefore you do not only create effort through repetitive segmentations, you also want to make sure every tool has the same information about the customer – impossible. A centralized solution for customer data management (collection, unification, insights, segmentation & activation) approaches this requirement – the system of insights.

Create new value

I was impressed as customers told me about the value that just being able to segment on data that was not available before, is bringing them. Data accessibility and enrichment allow them to address a bunch of new business cases, not only for the marketing department.

Thanks to normalization and transformation, users become able to use their data in a more structured way and create new connections between objects and data sources.

With unification you can identify an individual across different sources, even if their profiles do not share a master identifier. Unified profiles allow us to target individuals based on information that previously was not available for each touchpoint. In the same way CDP improves personalization based on reconciled data.

Having all the data ingested, modeled and unified, Data Cloud enables you to create new abstractions and aggregations with Calculated Insights. We did similar things in Marketing Cloud SQL Activities but with much less data being available. This is the functionality to do grading, scoring and to extend the utilizable information about your individual profiles.

All previous steps feed into segmentation. Giving the business users a segmentation interface where they can build audiences on their own without the need of IT (to prepare data and write SQL statements) can reduce the time to create audiences dramatically. Companies told me they are saving multiple days which they usually needed to set up data delivery and Automations for each campaign.

Every segment refresh then leads to an Activation to any system of engagement. Users can reuse segments and again save time they previously needed to build the segments in different tools.

No replacement for Marketing Cloud

Having Data Cloud in place I see customers reducing Marketing Cloud Automations as much as they can. Sometimes it’s still needed to do additional steps in Marketing Cloud before ingesting data into Journey Builder, but we can already see changes there. For the future I expect audience management in Marketing Cloud to reduce even more. What Data Cloud does not replace is the content management and the whole execution part of Marketing Cloud. Studios and Builders deliver way more value than just sending out a message to predefined audiences. The same applies to MC Connect. Being focussed on sharing customer segments, Data Cloud does not offer the same capabilities to access, utilize, create and update CRM data yet.

Conclusion

There are plenty things that make Marketing Cloud Engagement great that are not just replacable by a new tool. However if data and insights are the fuel of great experiences, Data Cloud is becoming to be the best petrol station for Marketing Cloud. It creates new opportunities for business and provides speed as well as scalability so that Marketing Cloud Engagement can focus again on its core purpose: delivering outstanding campaigns and 1:1 personalizations.

_{*This article is a private work and not an official Salesforce publication}