Case study · Sigma reporting on Snowflake

Turning a printed daily report into a report the desk can query

A commodity brokerage assembled its daily market report in a spreadsheet and sent it as a static sheet. Greyfield moved the numbers into dated Snowflake tables and rebuilt the report in Sigma. The desk can now open any day in the last two years, while refresh and distribution run on a schedule.

5 hrs to 30 min

a week producing the report

2 years

of history, any date reproducible

~1 month

to both daily reports live

Technology

Snowflake

Sigma

At a glance

IndustryCommodity brokerage

LocationNew York, USA

ChallengeThe daily report came out of one spreadsheet and went out as a static sheet.

EngagementReporting rebuild in Snowflake and Sigma

ResultFive hours a week of report assembly down to about thirty minutes

The challenge

The daily report came out of one spreadsheet and went out as a static sheet. Readers could see the day it was sent and nothing else. Comparing today against last Tuesday meant digging up an old file, and the whole process depended on one workbook and the person who maintained it.

Approach

What Greyfield did

Moved the numbers into the warehouse

Built report-serving tables in Snowflake so each report reads a dated table instead of formulas living inside a spreadsheet. The tables hold the history needed to reproduce any past day.

Rebuilt the report in Sigma

Recreated the printed layout in Sigma, then added a date picker over two years of history. The reports read Snowflake tables directly rather than custom SQL buried in the workbook, so reviewers can inspect the source logic.

Schedule refresh and distribution

Pipelines refresh hourly through the morning. The report goes out on a weekday schedule to the daily readers, with a weekly roll-up to the wider team.

Validated against the spreadsheet

Compared the rebuilt report against the original spreadsheet cell by cell, at the printed precision, so differences could be triaged as either a fix or a known data question rather than argued about. Validation continues as remaining differences are worked through.

Results

What changed for the client

The desk can open any report date from the last two years instead of hunting for an old file. The reporting logic now sits in Snowflake tables instead of one spreadsheet maintained by one person. Refresh and distribution run on a schedule. Both daily reports were live about a month into the engagement, with a weekly report following.