December 2025 Summaries

Customer engagement measurement often involves manual processes and fragmented data, but AI Columns offer an automated solution by integrating AI-driven insights directly into tools like Salesforce. By using Fivetran Activations to define datasets and leveraging large language models, AI Columns transform structured data such as product usage and customer attributes into actionable insights like engagement scores, summaries, and recommendations. The setup involves creating queries to pull relevant engagement data and using GPT-based prompts to generate insights. This integration allows sales, marketing, and RevOps teams to access enriched data without complex coding or maintenance, ultimately providing clear engagement scores and personalized talking points for account outreach.

Census Audience Hub, now part of Fivetran, provides marketing teams with the ability to access and utilize complete and trusted customer data directly from their data warehouses without needing engineering support. This platform allows marketers to create and sync audience segments across various marketing and advertising tools using a visual segment builder, eliminating the need for CSV uploads and complex SQL queries. Audience Hub offers over 200 pre-built connectors for seamless integration with marketing platforms and ensures that any changes made to audience data are updated automatically across all platforms. The tool promotes collaboration between marketing and data teams by making trusted data models self-service while maintaining data governance through an approval process and integrations with tools like dbt and Looker. Unlike traditional Customer Data Platforms, Audience Hub is warehouse-native, providing a unified and comprehensive view of customer data, enabling more effective decision-making and campaign personalization.

Reverse ETL is revolutionizing marketing by enabling teams to use high-quality, governed data directly in their everyday tools, thereby enhancing the precision and effectiveness of campaigns, experiments, and customer experiences. This technology allows for the seamless movement and activation of data across platforms, enabling marketers to deliver personalized messages, understand cross-channel conversions, and segment customers based on real-time engagement signals. Companies like Notion and Canva illustrate the transformative power of reverse ETL, as they have been able to improve customer segmentation, streamline workflows, and enhance the targeting and personalization of their marketing efforts. By integrating reverse ETL, Notion has simplified data sharing across platforms, while Canva has improved data usability for targeted messaging, allowing both companies to run more effective and timely marketing strategies. Overall, reverse ETL empowers marketing teams to maximize their impact by automating data processes, thus converting data warehouses into powerful growth engines.

Customer Data Platforms (CDPs) are being increasingly challenged by reverse Extract, Transform, Load (ETL) processes, which offer a more efficient and reliable means of utilizing customer data from data warehouses by directly integrating it into marketing and sales tools without creating unnecessary data copies. While CDPs suffer from rigid data models, long onboarding times, and redundancies, they also fail to store the comprehensive customer and company-level information available in data warehouses, which diminishes their claim as a "single source of truth." In contrast, reverse ETL empowers data teams to activate governed data at scale, facilitating better personalization and targeting for business teams without creating new data silos. Platforms like Census, now part of Fivetran, provide version control, logging, and observability, offering marketers a no-code interface for managing data segments and enabling smarter campaigns and measurable growth. As companies embrace the modern data stack, reverse ETL solutions are proving to be more agile and effective than traditional CDPs, delivering real customer insights directly to the tools where they can drive the most impact.

Data teams play a crucial role in modern businesses by providing insights essential for daily operations, yet they face challenges in demonstrating their impact due to issues with data quality and availability. Companies like Canva, Clearbit, and Loom have addressed these challenges by implementing reverse ETL, which allows data to be moved efficiently between different platforms, enabling more sophisticated use of data in business processes. Canva used reverse ETL to enhance data segmentation and personalization by integrating data from Snowflake into platforms like Braze and Salesforce, saving time and unlocking more value from their tools. Clearbit overcame data silos and redundant processes by using reverse ETL to sync insights from Redshift to Salesforce, allowing better customer views and personalized marketing efforts. Loom utilized reverse ETL to centralize customer data, enabling efficient ticket prioritization and enhancing communication with tools like Zendesk. These examples demonstrate how reverse ETL empowers data teams to operationalize data, streamline processes, and drive business growth by improving data accessibility and utilization.

November saw significant advancements across three key open-source projects: SQLMesh, SQLGlot, and the Fivetran Connector SDK. SQLMesh focused on enhancing reliability and compatibility, notably with its v0.209.4 release, which improved dbt compatibility and introduced MERGE support for the Fabric engine. SQLGlot made substantial progress with 125 merged pull requests aimed at enhancing dialect fidelity and type-aware analytics, particularly benefiting platforms like DuckDB and Snowflake. Meanwhile, the Fivetran Connector SDK experienced a hackathon-driven expansion with 30 new pull requests that contributed to a broad range of new connector examples and improved developer experiences. These projects reflect ongoing collaboration and innovation, with improvements enhancing capabilities for those involved in dbt projects, SQL transpiling, and custom data connectors. Looking forward, December is expected to continue these development themes, focusing on dbt and engine refinements, semantic and dialect advancements in SQLGlot, and further connector expansion and platform enhancements in the Fivetran Connector SDK.

Fivetran has introduced a new SQL Server binary log reader aimed at enhancing data ingestion for high-volume, performance-sensitive environments without disrupting day-to-day operations. This log-based replication method captures changes directly from the SQL Server transaction log, ensuring a low-impact, efficient transfer of data to warehouses or data lakes. It is particularly beneficial for organizations requiring minimal overhead on SQL Server instances, constant data updates, low latency, and scalability. This approach differs from traditional query-based methods by operating outside the database's query processing engine, thus maintaining production workloads unaffected and offering superior performance. The binary log reader also features automatic fallback capabilities, Always On Availability Groups support, and automatic schema migration, making it suitable for complex environments, whether on-premises, cloud-based, or hybrid. This tool provides a more efficient pathway for real-time analytics by enabling rapid, reliable data replication while keeping operational costs down and ensuring continued database performance.

Fivetran has been recognized for the sixth consecutive time as a Challenger in the Gartner Magic Quadrant for Data Integration Tools, reflecting the growing trust in its capabilities to support AI-driven futures. Over the past year, Fivetran has enhanced its offerings by expanding its Managed Data Lake Service, introducing the Connector SDK for custom data integration, and improving support for unstructured data. Through strategic acquisitions such as Census and Tobiko Data, Fivetran has bolstered its data movement, transformation, and activation capabilities, providing a secure and scalable platform for enterprises. Handling over 7,200 TB of data monthly across 700+ connectors, Fivetran supports over 7,600 customers worldwide, including leading companies like Saks and Autodesk, by delivering reliable pipelines that support analytics and AI at scale. The data landscape is experiencing significant changes, with a shift towards integrated and standards-based platforms that offer AI-ready data with speed and efficiency. This evolution is driven by the need to reduce tool sprawl, ensure reliable data access for AI and analytics, and respond to the trend of mergers and acquisitions, as enterprises look to simplify and automate processes. Fivetran aims to lead this shift by providing a versatile platform that supports interoperable data architectures, open formats, and the seamless activation of insights, ensuring enterprises can achieve more with less complexity.

Fivetran and dbt offer a powerful combination for streamlining data ingestion and transformation, enabling businesses to answer complex, cross-platform questions through natural-language queries without needing to write SQL or Python. Fivetran simplifies the replication of data from over 700 sources into centralized locations, while dbt transforms this raw data into clean, well-tested models, employing software engineering best practices such as CI/CD and modular development. This integration allows for the creation of a unified dataset optimized for analytics and AI, significantly reducing costs and improving performance. New AI-powered features like dbt Copilot and dbt Canvas enhance usability for non-engineers, while the Fusion engine improves execution speed and SQL comprehension. The combination of Fivetran's data replication, dbt's transformation capabilities, and AI-driven insights empowers users to conduct sophisticated analyses and make informed decisions across various platforms, ultimately transforming how organizations manage and utilize their data.

An end-to-end data integration architecture combines Fivetran, Snowflake Cortex, dbt, and Census to create a seamless workflow that transforms raw data into actionable intelligence, specifically tailored for ranch management operators. This setup enhances the typical data pipeline by integrating AI enrichment during data ingestion, transforming data with dbt without manual intervention, and delivering timely alerts via Slack through Census. The custom Fivetran connector is designed to enrich weather data with livestock health insights using Snowflake Cortex Agents before it lands in Snowflake, enabling real-time, proactive decision-making for ranch managers. The process begins with data extraction from a weather API, followed by AI-driven enrichment, automatic transformation into analytical views, and operational alerts sent directly to mobile devices, streamlining decision-making and reducing manual workload. This architecture exemplifies the potential of combining data ingestion, intelligence, transformation, and activation into a cohesive process, applicable to various industries needing timely and actionable data insights.

Product-led growth (PLG) is revolutionizing Revenue Operations (RevOps) by leveraging existing data to enhance operational tools and drive better outcomes, though it presents certain challenges. Experts like Henry Mizel, Cyril Marques, and Sylvain Giuliani emphasize the importance of a 360° view of product usage, enabling sales and customer success teams to prioritize accounts and prevent churn through authentic communication based on actual product interactions. Apollo's use of product data improved conversion rates significantly and facilitated automated upsell alerts, demonstrating the power of product-qualified lead (PQL) scoring in prioritizing sales leads. However, product data often remains siloed, prompting the need for a centralized data warehouse, as shown by Apollo's collaboration with Montreal Analytics to implement a modern data stack in Snowflake. This approach ensures deeper insights and confidence in data quality, which is crucial for actionable insights. Marques suggests an agile approach to data activation, starting small and focusing on key priorities, while emphasizing the importance of reliable data to maintain trust. PLG demands continuous experimentation and iteration, with tools like Census facilitating faster segmentation and audience testing. Ultimately, a solid foundation of trustworthy data allows organizations to be agile and effectively activate data across their business.

Over a 38-day period in 2025, Google Cloud, Elastic, and Fivetran hosted an AI-themed hackathon, challenging participants to create innovative products using Google Cloud tools and either Elastic's or Fivetran's capabilities. The event saw 2,770 participants and 171 project submissions, with winners receiving cash prizes and social media promotions. The first-place project, Alzora, aimed to assist patients with cognitive decline through a sophisticated data architecture utilizing Fivetran, BigQuery, and Vertex AI, offering features such as wearable data monitoring and GPS tracking. Second place went to DocuLytics, which transformed raw DocuSign data into actionable insights through automated analytics and a custom Fivetran connector, aiding legal and sales teams in compliance and customer analysis. MergeMind, the third-place project, automated code review tasks using a custom connector for GitLabs and dbt transformations to improve project management efficiency. The hackathon highlighted the significant role of AI in developing new capabilities at the application layer, leveraging curated datasets and advanced architectures to drive business innovation.

Canva has revolutionized its approach to customer data management by building a custom data stack to gain a comprehensive understanding of its users, which surpasses the capabilities of traditional customer data platforms (CDPs). This approach integrates best-in-class tools like Snowflake for data warehousing, Census for data activation, Fivetran for data ingestion, and dbt for data transformation, enabling Canva to consolidate customer information from various sources into a single source of truth. This architecture allows for enhanced audience segmentation and personalized marketing efforts without redundant data silos, thus optimizing customer engagement and conversion rates. By leveraging this system, Canva can efficiently target users with tailored content, enhancing its marketing strategy while reducing costs associated with traditional CRM solutions like Braze. This approach allows for more precise marketing attribution and identity resolution, while also providing greater flexibility, accuracy, and control over customer data.

Fivetran has introduced private networking for its Managed Data Lake Service, allowing enterprises to securely transfer data to cloud storage without using the public internet. This service combines the advantages of a data warehouse's query-readiness with the flexibility and cost-effectiveness of a data lake, providing a robust data foundation for decision-making, analysis, and AI/ML workloads. Private networking enhances security by keeping traffic on the cloud provider's internal backbone and restricting access to approved clients through bucket policies, while also improving speed and simplicity by utilizing native private connectivity features from major cloud providers like AWS, Azure, and Google Cloud. Designed as a premium feature for Business Critical plan users, this service supports large volumes of structured, semi-structured, and unstructured data from over 700 sources, offering a secure and efficient solution for data lake migrations or enhancing security postures.