Data Cloud April 2025 Release Highlights

This April, an impressive set of important updates are making their way to the platform that can improve usability and reduce operating costs. Certain Salesforce data sources can now be ingested for fewer credits, currency support is now GA including currency conversion, a new API is available to allow for real-time server-to-server ingestion, Zero Copy sources can now use file federation to reduce 3rd party data lake compute costs, and marketers can use vector search operators in their segmentation, allowing them the ability to search for terms of semantic similarity rather than having to get the exact keywords correct. Read on for more details.

• Internal Data Pipeline Credit Lever: Starting April 15, structured data ingested from CRM, Marketing Cloud, MC Personalization and Commerce Cloud will now be charged at a new rate of 500 credits per 1 million rows regardless of whether it’s batch or streaming, representing a 75% discount over the current rate for the same sources.
• Currency Data Type Support: Data Cloud now supports a native Currency data type, allowing businesses to ingest and store values in local currencies without requiring normalization at the point of ingestion. Each currency value consists of a numeric amount and a 3-letter ISO code, which can either be included in the source data or defined during data stream configuration. Currencies can be converted automatically at retrieval time for key functional areas like transforms, calculated insights, segmentation, and data actions, ensuring unified currency handling in filters, aggregations, and comparisons. Currency conversion rates are automatically brought in from all connected core CRM orgs, and the Data Cloud admin must select one as the default source for conversions across the platform—a choice that can be updated at any time.
• Real-Time Ingestion API: This new capability allows data ingested via the Ingestion API—not just SDKs—to flow directly into the real-time layer, enabling sub-second latency for use cases such as live credit scoring, vehicle telemetry, telco data alerts, and more.
• File Federation Support for Zero Copy: Currently in Beta, file federation is a new capability within the Zero Copy Federation suite that enables customers to read data directly from external cloud storage (like Snowflake, Databricks, or S3) without using those Data Lake's compute resources. Unlike Query Federation, which relies on customer-managed compute and can incur performance and cost challenges—especially with large datasets—File Federation offers a cheaper, faster, and more scalable approach by accessing Iceberg-format files directly. This significantly simplifies setup, reduces costs, and accelerates time to insights, making it ideal for organizations managing massive volumes of structured data.
• Vector Search Operator in Segmentation: This new Beta capability allows marketers to filter audiences based on unstructured text attributes using semantic similarity rather than exact keyword matching. The feature works by comparing vector embeddings using a dot product and similarity threshold defined by the user (e.g., 90%). This offers a major upgrade over traditional operators like is in, contains, or equals, which rely on exact terms or pattern matching rather than contextual meaning. With this, marketers can create more targeted and relevant segments from data that previously went untapped, improving campaign precision and ease of use. The capability is currently in beta, supports only inline unstructured text with pre-generated embeddings, and integrates into the segmentation UI via a new "is similar to" operator.
• New Connectors for Marketing Intelligence: Amazon Ads, DV360, Taboola, and Outbrain are now supported for the rapid ingestion and pipelining of data to be used within Salesforce's newest analytical tool, Marketing Intelligence
• Batch Transforms: Merge Mode: This new feature for batch transforms allows records to be upserted instead of replacing the entire output object, addressing inefficiencies with reprocessing previously transformed data. This is especially valuable for incremental data pipelines where only new or updated records need to be written, reducing compute and improving performance. Users can now choose between Replace (default) and Merge modes at the output node, giving them greater control and efficiency in their transform workflows.
• Hierarchical Aggregation in Segmentation: This feature allows marketers to roll up metrics like opportunity amounts or sales orders across account hierarchies, enabling more strategic targeting and insights. This is especially useful in B2B contexts, where parent-child account relationships are common. Users can now apply aggregation filters across related attributes using a new “hierarchical aggregation” checkbox in the segmentation builder. Up to 3 hierarchy levels and up to 5 filter containers using hierarchical aggregation are supported per segment.
• Communication Capping: this new feature enables businesses to control and optimize how often customers are contacted across channels and business lines, helping prevent message fatigue and overspending. Admins can configure custom dimensions like communication channels, business units, or customer segments (e.g., net worth tiers), and define daily or cumulative limits at both the individual and group level. During activation, these rules automatically apply, ensuring campaigns stay within defined thresholds, with clear visibility into which profiles were included or excluded based on the cap.
• Search Analytics: this feature provides visibility into how users interact with search within Salesforce, tracking queries, results, and engagement to help teams optimize search performance and content relevance. The feature leverages Data Cloud’s calculated insights and dashboards, surfacing metrics like most-searched articles, query success rates, and interactions that influence outcomes like case resolution. Organizations can use these insights to refine search experiences, reduce no-result queries, and improve self-service efficiency across channels.
• Data Kit Enhancements: Sequencing: Previously, users had to manually define the deployment order of components in a data kit, taking into account dependencies to avoid failures. This manual process was error-prone and time-consuming. With the new enhancement, the deployment sequence is automatically generated based on the components’ creation dates, reducing mistakes and improving reliability and efficiency.Relationship Packaging: Previously, DMO-to-DMO relationships only deployed to the default data space, requiring users to recreate them manually in non-default spaces. With the new enhancement, these relationships are automatically included in the data kit when both DMOs are present, and they deploy correctly to non-default data spaces. However, devops data kits can only deploy to a single specified data space, so standard data kits or managed packages should be used for deploying to multiple data spaces.