UD4D — Use data for decisions

Cutting VAT reporting compute costs by over 80% on Snowflake

We rebuilt a costly, hard-to-maintain reporting pipeline into a lean, modular Snowflake setup that is cheaper and faster to run, with identical outputs.

Get in touch Download PDF
Cutting VAT reporting compute costs by over 80% on Snowflake

About the client

Global pharmaceutical company running centralized VAT compliance reporting across multiple tax regions.

The challenge

VAT compliance reporting ran on a legacy Snowflake architecture of stored procedures and temporary tables that were costly to run and hard to maintain.

  • Monolithic stored procedure running the main reporting flow, hard to maintain and scale
  • Wide SAP S/4HANA source tables pushing large volumes of unused data through every run
  • Heavy use of temporary tables inflating runtime and obscuring the transformation logic
  • Oversized Gen1 warehouse driving high compute cost, with limited visibility into processing

The solution

UD4D refactored the pipeline in two phases, reducing cost and runtime without changing reporting outputs.

  • Code and query review of the stored procedure, simplifying the most expensive transformation steps
  • Removal of 11 unused source tables and 3,869 unused CRT stage columns, roughly an 88% reduction of columns at that layer
  • Replacement of stored procedure logic with modular, parallel Snowflake Dynamic Tables, and removal of all temporary tables
  • Filtering moved upstream, plus removal of around 60 of 150 reporting columns
  • Documentation of 100+ objects and their columns (~2,500 columns) and right-sizing to an XS (Gen2) warehouse

Business impact

  • 83%

    lower Snowflake compute

    Weekly usage fell from around 61 credits to about 10, an estimated 2,600 credits and roughly $9K saved per year.

  • 7x

    faster reporting runs

    Execution dropped from 40 to 50 minutes to 6 to 7 minutes through parallel processing and simplified logic.

  • 88%

    smaller processing footprint

    Removing 3,869 unused CRT stage columns and redundant reporting columns cut the data handled on every run.

  • 100+

    objects fully documented

    Table and column-level comments across roughly 2,500 columns replaced an opaque monolith with a modular, traceable pipeline.

Frequently asked questions

What was the problem?

VAT compliance reporting ran on a legacy Snowflake architecture of stored procedures and temporary tables that was costly to run and hard to maintain.

How did UD4D refactor it?

In two phases: reviewing and simplifying the stored-procedure logic, removing unused source tables and columns, replacing the stored procedure with modular, parallel Snowflake Dynamic Tables, and right-sizing the warehouse, without changing reporting outputs.

Which technologies were used?

Snowflake with Dynamic Tables, SAP S/4HANA sources and Informatica IICS.

What were the results?

About 83% lower Snowflake compute, roughly 7x faster reporting runs, an 88% smaller processing footprint, and 100+ objects fully documented.

Related case studies