Top results
Data Vault development dropped from hours to minutes: Templated build logic replaced hand-coding. Engineers now spend their time fine-tuning and building new data products instead of rebuilding pipelines, and they expect another jump as V2 Nodes reach more complex queries.
Teams are delivering roughly 3x faster, and they're still ramping up: Even mid-adoption, they're shipping about three times the work with the same headcount. VodafoneZiggo expects that number to climb as the migration finishes.
WhereScape retired in under a year, with about 2,000 PowerCenter objects migrated so far: VodafoneZiggo lifted and shifted its entire WhereScape environment onto Coalesce within 12 months. Roughly 2,000 of some 11,000 PowerCenter objects have moved across, and each one retires another piece of end-of-service tooling along with the cost and risk that come with it.
Development opened up beyond engineers: More and more analysts model directly in Coalesce, and build as they go. That leaves developers to fine-tune instead of rebuild, and it cuts repetitive work for everyone.
One change reaches everywhere through templates: Add a feature to a template once and every node inherits it. A better way of working spreads to every squad, and no one has to rebuild anything to get it.
Pain points
PowerCenter was out of sale and nearing the end of service. The transformation layer sat on Informatica PowerCenter, which had reached end-of-sale, an unsustainable foundation for a warehouse of ~1,500 workflows and 11,000+ target tables, prompting a push toward a decision.
The ETL layer couldn’t exploit Snowflake’s performance. After migrating to Snowflake, the existing PowerCenter pipelines couldn’t tap the platform’s processing power, leaving performance and cost savings on the table.
Pushing logic into Snowflake meant manual SQL override coding. Wherever the team moved transformation logic onto Snowflake, PowerCenter’s overrides forced hand-written SQL — slow to build and hard to keep consistent.
High development times and underperforming workflows at scale. Across ~1,482 workflows, ~9,447 sessions, and ~11,388 target tables, slow development and poorly performing jobs had become a structural bottleneck.
The challenge
VodafoneZiggo faced a deadline it couldn’t miss. With PowerCenter nearing the end of service and a warehouse that size to keep running, the data team needs to replace the transformation layer without re-platforming the data model or stalling delivery.
VodafoneZiggo wanted more than a like-for-like swap. It set out to find a platform that used Snowflake directly, automated the repetitive coding, and fit its existing Data Vault model on day one.
The evaluation
VodafoneZiggo weighed candidates against a short list of hard requirements.
Key criteria for evaluation included:
- Use Snowflake’s power directly: an ELT platform that pushes processing into Snowflake rather than working around it.
- Code automation and deployments: less manual coding, with rapid development and deployment across all data layers — driving down time to market.
- Standardization through template customization: the Scalefree Data Vault package prooved to be adaptable to VodafoneZiggo’s internal DV standards, and provide immediate compatibility with the existing data model.
- Column-level lineage: lineage down to the column, including the code snippet, with the ability to propagate.
- Open metadata: insights into data quality and operations.
- Immediate fit with the existing Data Vault model: no re-platforming required to adopt the platform.
Coalesce met every criterion, and VodafoneZiggo ran a proof of concept in 2023 before committing.
The migration
VodafoneZiggo began with the Scalefree Data Vault package for Coalesce and adapted it to its internal Data Vault 2.0 standards. Because the package already matched the model, none of the existing work had to be redone. The data team ran the training alongside Coalesce, putting more than 50 people through four in-house sessions and a hands-on lab. The first pipelines hit production in July 2024, and from there the rollout ran almost daily on Git-based CI/CD and Airflow orchestration.
As it was built, VodafoneZiggo automated every repetitive step it could find, from key conversion based on naming conventions to bulk node edits. That habit of questioning each manual action produced nine custom templates aimed squarely at speeding development. By running Coalesce on an XS Snowflake warehouse, where many PowerCenter processes had needed an M connection, VodafoneZiggo lowered compute costs without sacrificing throughput. Column-level lineage, complete with code snippets and propagation, gave the data team confidence to change logic in one place.
The impact
The payoff came fast. VodafoneZiggo cut Data Vault development from hours to minutes, brought more roles into building, and retired an end-of-service platform on schedule. Most of that happened before teams had even finished learning the platform.
Templating Data Vault 2.0 into nine reusable build patterns
Before templates, every Data Vault object meant hand-coding the same patterns over and over, one of the biggest drags on delivery in a warehouse this size. VodafoneZiggo turned that repetition into nine reusable build patterns that generate logic straight from metadata. Work that once took a data engineer hours now takes minutes, and every pattern carries VodafoneZiggo’s own Data Vault 2.0 standards. The Scalefree package gave the team a head start that already fit the model, so there was nothing to redesign.
With the build logic templated, data engineers stopped rebuilding pipelines and instead started fine-tuning them and building new data products. The shift resets what a data engineer’s day looks like: less boilerplate, more design. Standardization came along for free because a single template applies the same naming, key generation, and structure to every node built from it, and automated key conversion based on naming conventions removed a manual step that had been easy to get wrong. VodafoneZiggo expects a further jump in speed as V2 Nodes reach their more complex queries.
Automating repetitive work and opening development to more roles
The gains reached past the data engineers. With more analysts now building in Coalesce themselves, modeling and constructing is now done in one step instead of handing specs off and waiting on engineering. That removed a familiar bottleneck: analysts used to model the data, then wait for someone else to build it. Repetitive work fell across the whole team. And even with most people still learning the platform, VodafoneZiggo is already delivering about 3x faster, with more headroom as the migration wraps up and the team builds experience.
Retiring legacy ETL and scaling standardization with AI
VodafoneZiggo lifted and shifted its entire WhereScape environment onto Coalesce inside 12 months, retiring the cost and risk of that legacy platform. The PowerCenter migration is further along: about 2,000 of its roughly 11,000 objects have moved, with the rest queued, and VodafoneZiggo retires end-of-service tooling piece by piece as it goes.
A key driver of this success has been the creation of a central platform squad, tasked with maintaining the standardized templates used across all pipelines. This team ensures consistency, scalability, and efficiency in development by delivering automated solutions and reusable patterns. By centralizing template management, feature squads can focus on development while relying on the central squad for robust automation and design principles. This collaborative approach has enabled VodafoneZiggo to scale its Data Vault architecture efficiently and accelerate delivery timelines significantly.
Templates keep standardization spreading. When a squad needs a new feature, VodafoneZiggo Coalesce Platform squad adds it once to a template, and every node built from it gains the capability, so a better way of working reaches every team without rework. Next, VodafoneZiggo plans to use AI to accelerate the remaining PowerCenter migration and is running a fast-track project for Holland’s Nieuwe Fixed launch, moving prototype data marts into a governed Data Vault track.
