viadukt

When data accuracy is business critical

viadukt is a fast growing German energy-efficient renovation platform, connecting homeowners and institutions with the consultants, tradespeople, and government funding necessary to transform properties for the energy transition. With 279 million building parameters powering their comprehensive data set, they enable stakeholders across Germany's €185 billion renovation market to make informed decisions about sustainable building practices.

For viadukt, data accuracy isn't a nice to have, it's a core part of the product and value-offering. When data quality issues began compromising their platform's analyses and recommendations, customer complaints threatened to undermine the trust that positions viadukt as the go-to partner for energy-efficient renovations.

Scaling trust across Germany's building stock

viadukt operates at the intersection of Germany's ambitious climate goals and digital transformation. 35% of Germany's energy consumption is from residential and commercial buildings, so eco-friendly renovations are necessary if Germany is going to meet it's climate neutral goals by 2045. viadukt's platform serves as the digital gateway for these renovation projects (Source.)

The company scaled systematically: starting with Wuppertal (their home base and initial customer region), then expanding to the larger state of NRW, and finally scaling to cover all of Germany. However, complaints began to surface significantly after the nationwide rollout, as the team struggled to ensure accuracy of their millions of new data points.

viadukt's platform provides:

• Instant building analysis from just an address input
• Renovation recommendations with cost simulations and ROI calculations - delivering precise results automatically, with no additional user input required.
• Government subsidy optimization connecting to billions in available funding
• Marketplace integration linking property owners with certified craftspeople, energy consultants, as well as financial institutes.

The data accuracy crisis

viadukt's most critical data accuracy challenges centered on building footprints and elevation (number of levels). These two factors directly determine the net heated area, a foundational input for all renovation simulations.

Building Footprint Errors: Inaccurate or mismatched spatial data sometimes linked the wrong footprint to a property, or even to the wrong address entirely. This led to fundamentally flawed modernization assessments.

Elevation Data Errors: Incorrect estimates of a building’s number of levels compounded the problem, as net area is calculated by multiplying the footprint by the number of levels.

Impact on Simulations: Even small errors in footprint or elevation could cause net area estimates to deviate by over 30%, severely skewing energy consumption, cost simulations, and ROI calculations.

Unsustainable workflows and increased business risk

Customer complaints about incorrect building assessments were becoming a persistent problem. The three-person data team found themselves in triage mode. They were manually investigating reports of inaccurate data, tracing root causes across complex data pipelines, and then deploying fixes.

Their workflow was unsustainable and reactive:

• Manual investigation of each complaint took up over 80% of the team's workweek
• Errors were found only after customers encountered them
• Time-intensive debugging across massive parameter datasets

Each incorrect analysis could cascade through viadukt's ecosystem, affecting renovation recommendations, loan approvals from partner banks, and relationships with the craftspeople marketplace. The complaints were particularly damaging when they came from large customers managing extensive portfolios, directly undermining trust and leading to significant losses in modernization orders.

Finding data validation at scale with Recce

The breakthrough came when team member Alexey discovered Recce through an article shared by a friend, which highlighted Recce's use case. Interested, he pitched it to the team, and they began evaluating how Recce could integrate with their stack. After a brief testing phase using Recce's OSS and connecting with the Recce team, the collaboration took off.

The requirements were specific:

• Systematic validation across massive building datasets
• Change impact visibility when updating data models
• Collaborative review process for the distributed team
• Integration with existing dbt and CI/CD workflows

From local script to cloud collaboration

Phase 1: local validation with OSS

The team started with Recce's open source software and local implementation. Alexey created scripts to prepare dev and production environments, enabling the team to:

• Visualize data changes through lineage and model diffs
• Focus validation efforts on modified columns and affected downstream models
• Document test results by copying diff outputs into PR comments
• Catch issues before they reached production

Recce's impact was immediate. The team could see exactly what changed and validate those changes systematically using features like lineage diff, schema diff, query diff, and value diff.

Phase 2: cloud transformation

As Recce Cloud became available, viadukt upgraded their workflow. Using Recce's OSS project, they had to build the base artifact for every review, which took 30 minutes every time. Now, with the cloud integration, they can simply start the review server and the correct state is automatically loaded, no manual building or downloading required.

Additional benefits included:

• Automated checks: Preset validations (row counts, schema diffs) caught common issues automatically across every PR
• Streamlined review: PR reviewers could access validation results directly without environment preparation
• Enhanced collaboration: Team members could access validation results directly in cloud-hosted environments without local setup
• IDE friendly: Enhanced developer productivity by providing an easy way to spin up a Recce instance for any specific branch

Rebuilding the trust

The outcomes transformed both the team's workflow and business impact:

• 70% reduction in customer data complaints
• Reduced data team firefighting from over 80% of capacity to less than 15%
• Proactive issue detection replacing reactive firefighting
• Faster review cycles with systematic validation evidence
• Increased team confidence in data pipeline changes

The biggest impact was cultural. The data team evolved from reactive firefighters to proactive data quality guardians:

• Systematic validation became standard practice
• Data impact visibility informed better development decisions
• Collaborative review process improved team knowledge sharing
• Confidence in deployments reduced stress and improved velocity

Improved data quality strengthened viadukt's entire platform:

• Customer trust restored and enhanced
• Partner confidence in data accuracy for loan decisions
• Platform credibility supporting business growth
• Competitive advantage through reliable building analysis

Data quality as competitive moat

viadukt's transformation illustrates a broader principle: for modern businesses, data quality isn't just a technical concern, it's a strategic differentiator.

For viadukt specifically:

• Accurate building analysis enables confident renovation decisions
• Reliable data supports partnerships with banks and craftspeople
• Trust in the platform drives marketplace network effects
• Quality at scale becomes increasingly difficult for competitors to replicate

As Germany accelerates toward its 2045 climate neutrality goals, platforms like viadukt become critical infrastructure. The companies that can deliver accurate, reliable building data at scale will capture the largest share of the massive renovation market.

Lessons for data-driven platforms

• Be proactive, not reactive: Systematic data validation prevents customer-facing issues more effectively than post-incident firefighting.
• Visibility drives confidence: Understanding data change impact enables teams to deploy with confidence rather than anxiety.
• Collaboration scales quality: Cloud-based validation tools enable distributed teams to maintain high standards by enhancing knowledge sharing
• Business context matters: Data quality investments should align with business criticality. Viadukt's data is directly visible to customers, this justifies higher validation standards than internal analytics.
• Cultural transformation: Improved processes changed the team's culture from reactive to proactive quality assurance.

About viadukt

viadukt is a fast growing german platform for building energy renovation, connecting property owners with financing, energy consultation, and certified craftspeople through their comprehensive datahub of 279 million building parameters. Founded in 2022 in Wuppertal, the company is driving Germany's building energy transition through digital innovation.
Learn more: viadukt.de

About Recce

Recce provides data validation and change impact assessment for modern data teams. By enabling systematic data diffing and collaborative review processes, Recce helps teams prevent data quality issues before they reach production.
Learn more: reccehq.com

Ready to transform your data quality process? Check out our docs to get started, or book some time directly with our team to learn how we can solve your specific problems. Join teams like viadukt and stop merging to production in the dark, merge with confidence after performing Recce.