empes

Case study · 2025

Reducing reporting friction through automation and data restructuring

A mid-size operations team spent hours every week building reports by hand that were out of date on arrival. We simplified the process first, then automated it, and the reports now update themselves.

Client
Operations / B2B services
Status
Shipped
Services
Empes Synthesis, Empes Infrastructure
Scope
Data engineering, automation, process improvement

Problem

Every week, someone exported data by hand, copied it across three separate tools and pasted it into a shared spreadsheet. The spreadsheet was often out of date, or did not agree with itself. Producing the reports took the team several hours each week, and by the time they were shared they were already stale.

So decisions were being made on data that was days old, and nobody fully trusted the numbers.

Approach

We started with a process audit. We mapped exactly how data moved from the source systems to the final report, and where errors, delays and manual steps came in.

The aim was not to automate the existing process. It was to redesign it first. Several steps were removed entirely. The rest were reorganised around a cleaner data model, and only then did we automate anything.

The weekly report, before and afterTwo columns. Before: a manual export, copy and paste across three tools, a shared spreadsheet that was often stale, and a report already days old. After: an automated pipeline, a restructured data model, and a report that updates itself, with alerts if something breaks.BeforeSource systemsWeekly manual exportCopy and paste across 3 toolsShared spreadsheet, often out of dateReport, already days oldSeveral hours a week, and nobody fully trustedthe numbers.AfterSource systemsAutomated pipelineRestructured data model: one source oftruthReport that updates itself, in the team'sformatIf something breaks, the team knows straightaway.audit,remove,redesign,thenautomateThe weekly report, before and afterTwo columns. Before: a manual export, copy and paste across three tools, a shared spreadsheet that was often stale, and a report already days old. After: an automated pipeline, a restructured data model, and a report that updates itself, with alerts if something breaks.BeforeSource systemsWeekly manual exportCopy and paste across 3 toolsShared spreadsheet, often out of dateReport, already days oldSeveral hours a week, and nobody fully trusted thenumbers.audit, remove, redesign,then automateAfterSource systemsAutomated pipelineRestructured data model: one source of truthReport that updates itself, in the team's formatIf something breaks, the team knows straight away.
Read the diagram as text
  1. Before: Source systems, then Weekly manual export, then Copy and paste across 3 tools, then Shared spreadsheet, often out of date, then Report, already days old. Several hours a week, and nobody fully trusted the numbers.
  2. audit, remove, redesign, then automate
  3. After: Source systems, then Automated pipeline, then Restructured data model: one source of truth, then Report that updates itself, in the team's format. If something breaks, the team knows straight away.

The method, with a payback calculator, is written up in the guide Reporting automation: cost, architecture and a worked example.

Implementation

Six work packages, in orderSix steps: process audit, data model redesign, automation pipeline, reporting layer, error handling and alerting, documentation and handoff.Process auditthe whole reporting workflow,mappedData model redesignsource data made consistentand reliableAutomation pipelinesources connected, manualsteps removedReporting layerclean output that updatesitself, in the team's formatError handling andalertingthe team knows at once ifsomething breaksDocumentation and handoffa full operational guide forthe teamSix work packages, in orderSix steps: process audit, data model redesign, automation pipeline, reporting layer, error handling and alerting, documentation and handoff.Process auditthe whole reporting workflow, mappedData model redesignsource data made consistent and reliableAutomation pipelinesources connected, manual steps removedReporting layerclean output that updates itself, in the team's formatError handling and alertingthe team knows at once if something breaksDocumentation and handoffa full operational guide for the team
Read the diagram as text
  1. Process audit (the whole reporting workflow, mapped)
  2. Data model redesign (source data made consistent and reliable)
  3. Automation pipeline (sources connected, manual steps removed)
  4. Reporting layer (clean output that updates itself, in the team's format)
  5. Error handling and alerting (the team knows at once if something breaks)
  6. Documentation and handoff (a full operational guide for the team)
  • Process audit. A full map of the reporting workflow as it ran.
  • Data model redesign. Source data restructured so it is consistent and reliable.
  • Automation pipeline. The source systems connected, and the manual steps gone.
  • Reporting layer. Clean output that updates itself, in the format the team preferred.
  • Error handling and alerting. If something breaks, the team knows immediately.
  • Documentation and handoff. A full operational guide, so the team can run it themselves.

Outcome

of manual work removed every week, approximately
4 to 6 h
  • Approximately 4 to 6 hours of manual work a week eliminated.
  • Reports update automatically and are always current.
  • Data consistency improved, with one source of truth across the team.
  • The team regained confidence in its numbers and started using data more actively in decisions.

What this shows

Automating a broken process makes it break faster. Simplify first, then build. The most valuable step in an automation project is understanding and simplifying the process, and then building the pipeline around something that works.

If one of your reports eats a day a week, book a call and bring it along.

Stack
Power Automate, Power Flows, Python
Type
Strategy + implementation
  • data-engineering
  • automation
  • reporting
  • process-audit
  • no-code
  • pipeline
  • operations

Book a 30-min scoping call

Bring one process. We will tell you whether automation changes it, and what it would take.

Book a 30-min scoping call hello@empes.it