Proof
Results you can check
Every figure on this site comes from delivered estates and is counted under the same rules, set out on the Methodology page. Customers are described, not named, until each gives written permission; their specific figures are added as permission lands.
Key facts
- Typically 95% or more of objects converted with no manual edit, under the one-line-edit rule.
- Reconciliation typically above 95% on first pass.
- 4 to 8 weeks on delivered estates of 1,000 to 1,600 objects.
- About 40% cheaper than an SI-led rewrite, at a fixed price.
- Three Databricks-commissioned enterprise migrations delivered.
Novuz figures are typical results on delivered estates, counted under the Methodology rules.
Novuz talk
Can you prove it?
A short talk on the question behind every figure on this page: can you prove the new version matches the old one? Same inputs, same outputs, a reconciliation report for every object.
Read the transcript
Can you prove it? It is the question every regulated data team now faces.
Data migration is solved. Logic migration is not. Up to forty percent of IT budgets now goes to technical debt. Most will be architectural by 2027.
Renewals and end-of-support dates turn that debt into a deadline. And the code that holds your rules is usually undocumented.
AI made conversion cheap, and proof harder. Hand rewrites are slow. LLM converters can change between runs. Regulated teams need evidence for every object, not a demo.
Your auditor will ask five things. Same input, same output? Every change explained? Evidence per object? Did anything leave your environment? Was the price known up front?
Three ways to convert logic. Novuz is the one that does all six: the same output every time, every change explained, evidence per object, code in your environment, the system deployed, and a fixed price first.
A free SQL translator is a good start. Novuz covers the rest: SSIS, BTEQ control flow, orchestration, and a reconciliation report for every object.
Modernization is not a one-time project. It is a capability you keep. Each cycle keeps your standards and system knowledge, so the next one costs less.
Deterministic modernization, ready for your agents. They drive Novuz by API and run its pipelines through MCP, under your controls. Same input, same output.
Start with your own code. A $3,000 assessment sets a fixed price in three business days. Or try the free scorecard, novuz.com.
The figures
What delivered estates typically show
- The counting rule comes first. An object counts as automated only if it needs no manual edit. A one-line fix counts as manual, and dead code is excluded from the count.
- Typically 95% or more of objects converted with no manual edit.
- Reconciliation above 95% on first pass; the report names the rest.
- 4 to 8 weeks on estates of 1,000 to 1,600 objects, from code received to reconciled output.
- About 95% of objects convert by fixed rules and the rest go to the resolver. That is the rule share, not the automation rate.
- About 40% cheaper than an SI-led rewrite, at a fixed price set before work starts.
Case study
Media analytics company, Snowflake to Databricks
The customer moved its Snowflake estate, including complex stored procedures across several databases, to Databricks. Every object was reconciled against its Snowflake output on the same inputs before cutover.
- Source: Snowflake · Target: Databricks · Outcome: full estate live on Databricks, every object reconciled
“Very happy with the high level of automation and validation. Our team was pretty hands off.”
Case study
A legal news and information publisher, Snowflake to Databricks
The publisher moved an estate of complex stored procedures and its data to Databricks, with a reconciliation report for every object.
- Source: Snowflake · Target: Databricks · Outcome: estate live on Databricks, reconciled object by object
“The migration was completed on time with minimal disruption to our operations. Strongly recommend Novuz.”
Case study
Advertising company, Snowflake to Databricks
Machine-learning pipelines and their data were moved to Databricks and went straight into production.
- Source: Snowflake · Target: Databricks · Outcome: pipelines in production on Databricks
Case study
A Fortune 500 natural gas utility, SQL Server to PySpark and Iceberg
Complex T-SQL stored procedures with recursive joins and subqueries were converted to PySpark. The utility is now modernizing its wider SQL Server estate to open Iceberg tables, with Novuz running inside its own environment.
- Source: SQL Server · Target: PySpark on Iceberg · Deployment: inside the customer’s environment
“This was a complete conversion and met our needs.”
Case study
An AI-powered supply chain planning software company, SQL Server to Delta Lake
The proof of concept covered the enterprise data warehouse and orchestration. Novuz was then deployed inside the customer’s environment for a self-service migration run by its own team. The program covers 8,000 workflows.¹
- Source: SQL Server · Target: Delta Lake · Scope: enterprise data warehouse and orchestration · Model: self-service, inside the customer’s environment
Partners
Who has validated Novuz?
Databricks validated technology partner. Three Databricks-commissioned enterprise migrations delivered.
Novuz is Virtusa’s digital solutions partner, enabling scalable, product-led enterprise modernization through joint collaboration.
Methodology
How are these results measured?
We use the same definitions on every engagement.
- Automation rate: the share of objects converted with no manual edits.
- Reconciliation accuracy: the share of objects whose outputs match the original on the same inputs.
- Elapsed time: code received to reconciled output, in business days.
- Object: one table, view, function, procedure, script, package or job.
Sample reports
What does a reconciliation report look like?
See exactly what your auditor will see, object by object, before you talk to us. Enter your work email and we send the redacted PDF.
“An auditor does not accept probabilities. Every object we convert ships with a report that shows it matches the original on the same inputs.”
- 1 An AI-powered supply chain planning software company, SQL Server to Delta Lake, self-service rollout. All figures are defined on the Methodology page.
Sources and further reading
- McKinsey & Company, "Tech debt: Reclaiming tech equity," October 2020: CIOs estimate tech debt at 20 to 40 percent of the value of their technology estate, and 60 percent say it has risen
- Gartner, Magic Quadrant for Technical Debt Management Tools, May 2026: by 2027, 80% of technical debt will be architectural
Written by Ted Ghose, Co-founder and CTO, Novuz
Last reviewed:
Find out what your logic debt is costing you.
Every object counted, lineage, dead code and a fixed price in three business days. $3,000, credited.
Not sure which door? The scorecard takes ten minutes and tells you.