Zendesk empowers GTM teams with customer insights and accelerates AI data readiness

“Fivetran, Snowflake, and dbt together have changed how every function in the business runs. Sales catches renewal risk before it becomes churn, finance reconciles revenue in hours instead of days, and marketing acts on campaign data while it still matters. That's the real return on this investment.”
— Akshay Agrawal, Director of Data Engineering at Zendesk

As the leader in AI-powered service, Zendesk helps businesses deliver faster, smarter, and more personal customer and employee experiences. Zendesk's platform is purpose-built for service — combining AI agents, automation, and human insight to make every interaction seamless. 

However, as the company scaled, access to enterprise data became critical across every function — from go-to-market (GTM) teams streamlining sales execution and customer retention, to marketing teams optimizing campaigns, to finance and operations driving reporting and planning. Furthermore, the unified data was required to support enterprise ML initiatives. Supporting these efforts required bringing together data from dozens of core systems and applications, including Salesforce, Marketo, Gainsight, Workday, and Zuora.

To meet this need, Zendesk’s data engineering team built a centralized data platform designed to serve as a single source of truth. However, ingestion remained a manual, engineering-intensive process. Each new data source required interpreting API documentation, developing pipelines, and continuously maintaining them as schemas evolved. As a result, enterprise data and product telemetry were siloed, making it difficult to generate a complete view of customer behavior.

This created a growing gap between how quickly the business needed data and how fast the team could deliver it. Engineering capacity was consumed by pipeline development and maintenance, while business units often waited months to access data. 

A data foundation built on Fivetran, Snowflake, and dbt

To close this gap, Zendesk rebuilt its data platform around Fivetran, Snowflake, and dbt.

Fivetran standardized data ingestion across the platform, replacing custom-built pipelines with a fully managed solution. Instead of building and maintaining DIY pipelines, the team can now onboard new data sources quickly and more reliably. Schema changes, historical data syncs, and API versioning are handled automatically — significantly reducing the operational burden on engineering and accelerating access to analytics-ready data. The team also benefits from Fivetran’s interoperability with Terraform and Airflow, as well as its ability to readily support multiple environments.

Snowflake serves as the centralized data platform, storing both raw and transformed data in a single governed environment. This ensures that departments across the business work from the same underlying data. 

The team uses dbt to model raw data into reusable, domain-specific datasets and define shared business logic across the organization. Version control and CI/CD for transformations allow teams to collaborate on data models and maintain semantic consistency. dbt is also used to track how data changes over time, enabling more accurate reporting and historical analysis.

‍

The impact: Faster insights, greater scale, and broader access 

New data sources that previously took 2 months to onboard are now available in ~2-3 weeks, allowing teams to access and validate data much earlier and translating into faster, more confident business decisions.

“With Fivetran and dbt, what used to take months now happens in weeks, giving the business faster access to trusted data and creating the foundation we need to scale analytics, agents, and AI across the enterprise.”
— Akshay Agrawal, Director of Data Engineering at Zendesk

Business lines across the company have seen considerable impact from unified, trusted data: 

• Go-to-market teams gained a 360° view of account health by combining Salesforce, Gainsight NPS, Gong calls, Zuora MRR, and product usage data, refreshed daily in Snowflake.
• Customer teams identified risk and expansion signal earlier with proactive visibility into churn, contraction, and opportunity indicators.
• Product and customer teams gained better product gaps and upsell signals by joining Gong data with support data, letting customers tell the teams exactly what is and isn’t working, at scale.  
• Finance teams accelerated revenue reconciliation by reconciling Zuora billing and Salesforce pipeline bookings in hours instead of days.
• Executives gained book-of-business visibility with an account research dashboard that brings together GTM signals in Snowflake.

This shift also changed how the data engineering team operates. Engineers who were previously focused on building and maintaining pipelines are now focused on more impactful projects like modeling data, defining metrics, and building AI-ready data foundations.

Across the broader organization, increased data access has reduced the ad hoc, shadow analytics that previously bypassed official processes and produced conflicting data models. The team is exploring new use cases with Fivetran involving increasing data coverage, reducing data latency, and using AI to analyze customer interaction transcripts and extract trends and intent.

[CTA_MODULE]