Automating a half-day reporting workflow
A weekly Excel process pulled together warehouse queries, file-based inputs and reporting outputs across roughly 20 working sheets. I started rebuilding it as a consolidated SQL + Python pipeline.
Process
The final report was much smaller than the work required to build it.
The weekly workbook combined multiple warehouse queries, Excel inputs and other reporting sources into season-to-date, monthly and overview outputs. The working macro-enabled file could reach roughly 50 MB, and automatic formula calculation could make it effectively unusable while Excel continued processing.
Under normal conditions the process took about half a day. If an upstream discrepancy appeared, troubleshooting could consume most of the day.
Before
Excel as the processing engine
Multiple query exports
~20 feed/working sheets
Formulas and lookups
Manual splitting and formatting
After
SQL + Python pipeline
Consolidated warehouse query
Remaining file inputs
Programmatic calculations
Generated business-facing workbooks
Redesign
I moved computation out of the workbook.
I consolidated the warehouse feeds into a broader SQL transformation and used Python to merge the remaining file-based inputs, apply additional classifications and calculations, and build the output.
The intent was not to remove Excel from the business. Excel remained the familiar deliverable; it just stopped carrying as much of the processing burden.
Output
The automation still had to produce a usable workbook.
The final reporting package kept the season-to-date, monthly and overview views people expected, and the workflow also handled founder-specific output generation. Formatting, formulas or values, merged headers, row heights and other workbook details mattered because the output still had to be ready to use.
Status
The project moved a fragile manual process toward a repeatable pipeline.
The SQL consolidation and workbook-generation pieces established the new workflow. I keep the impact claim intentionally conservative here because end-to-end production validation was still in progress during the implementation period.