Skip to main content

Health / Insurance · Automation & RPA

Automating three back-office processes and turning down two others

Five automation candidates, timed on site. Three cleared the ROI threshold, two did not. We documented why and built only the three.

Sector
Health / Insurance
Size
600 employees
Service
Automation & RPA
Model
Assessment then fixed prices
Duration
2-week assessment, 3 robots in 11 weeks

Context

The back office manually re-keyed data between the member management tool, a partner extranet and a control spreadsheet. Management had an automation budget and a list of five processes supplied by the business teams.

The problem

  • Manual re-keying between three systems with no interface
  • Effort estimates supplied by the teams, never measured
  • No API available on the partner extranet
  • Previous automation attempt abandoned after six months

What we did

  • Two weeks observing and timing the five processes as actually run
  • Gap between reported and real time ranging from -40% to +25% depending on the process
  • Three-year ROI calculation including maintenance, process by process
  • Two processes ruled out: too infrequent for one, business rules changing quarterly for the other
  • Three UiPath robots built with exception queues and named alerting
  • Four weeks running in parallel before cutover on each robot
  • Written and tested manual fallback procedure for each process

Results

3 / 5
processes retained
1 840 h
returned per year
9 mois
payback period
99,2 %
success rate at 6 months

What remains

The recovered hours were reallocated to handling complex files, not to headcount reduction — a condition set by management from the outset. The two rejected processes were covered by a note explaining why, which the client reused to arbitrate other internal requests.

« The most useful part of the assessment is the section explaining why two processes should not be automated. We still use it. »
Chief Operating Officer, Health insurance mutual

Stack

  • UiPath
  • Power Automate
  • SQL Server
  • REST API

Case studies

Twenty minutes is enough to know whether we are useful

No sales deck. You describe the need, we say whether it is in scope, at what price and on what timeline. If it is not for us, we say so during the call.