On paper processes are simple, linear, and known. Supply chain and inventory, finance, compliance, and product design, every department has its ways of working.
Systems hold the data
Plate Data, process modelling and value
Date 2026-10-03
Licence Rextuvia
Rextuvia finds the lost value
Field
On paper processes are simple, linear, and known. Supply chain and inventory, finance, compliance, and product design, every department has its ways of working.
Basis A simple purchase-to-pay procedure
Source Rextuvia’s drawing
Reality normally looks different. Deviations, workarounds, rework, approvals going back and forth, bottlenecks and delays are some of the many defects found in a process.
Basis The same procedure as systems record it; a schematic, no single company’s log
Source Rextuvia’s drawing
Systems keep a record of every action: which process, what activity and when it happened.
Process mining is a scientific field, begun in 1999. It compares the process as designed with the process as the systems recorded it.
Findings uncover defects that are traceable down to the record, so an organization can see where and how exactly it leaks value and money.
Basis Begun 1999; first journal paper 2004
Source W. M. P. van der Aalst et al., IEEE Transactions on Knowledge and Data Engineering, 2004
Field Evidence
Half
of the top 200 of the Forbes Global 2000 run process mining on Celonis, the largest of the platforms.
3 to 5%
of EBITDA lost in broken order-to-cash processes at one industrial manufacturer of about $10 billion in revenue, found by reading its own records.
$4.53
for every $1,000 of revenue is what the typical company spends running its purchasing: the people, the systems, the paperwork. The best-run companies spend $2.31, about half.
99%
of the cash discounts on offer now taken by Vetter Pharma, a drug manufacturer with €840 million of revenue in 2021, up from 80% once its own records showed which were missed: about €600,000 a year.
Processminingwithouttheplatform
Method
No licences, no training, no integrations to maintain.
One engagement report
Defect report. Where the process breaks down and why, each defect traced to the system’s records.
Value lost. What each defect costs: profit lost every year, profit recovered once, cash released once.
Recommendations. What to fix, defect by defect.
Purchase-to-pay and order-to-cash. The two processes that touch every supplier, every customer and every invoice.
Other processes a system records, such as inventory management, production, record-to-report or asset maintenance, can be added to a Project or to Monitoring, each quoted on effort.
The data comes from an enterprise resource planning (ERP) or internal operational system, under a non-disclosure agreement (NDA) and a data processing agreement.
Nothing to install. A data extract is transferred from SAP, NetSuite, JD Edwards, Microsoft Dynamics or other ERP systems.
Success depends on the quality of the data, so the framework starts with a Pre-audit: a free check of the systems, tables and fields, with no commitment.
Event log. The raw tables of the ERP system become one log: each order, each step, each timestamp.
Discovery, conformance and enhancement. The process is drawn from the log, held against the agreed reference model so every deviation is named, then enhanced with the times and volumes behind each path.
Value assessment. A price for each defect, agreed together and applied to every case found.
The report. One report in three parts: defect report, value lost and recommendations to recover the leaked value.
Pilot. Purchase-to-pay. 12 months of data. One site, the largest or most troubled.
Project. Purchase-to-pay and order-to-cash. 24 months of data: the last 12 measured, the 12 before them to show which defects are structural. Across every location on one ERP instance.
Monitoring. After the Project, the same scope, repeated every six months or once a year on new data.
Everythinghasprovenance
Provenance
“It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts.”
Every finding can be verified. Rextuvia works on the same logic, with the marginalia as its indicator: each finding traces back to the record it came from, the reasoning that counted it and the value agreed with the organization.
Estevão Castro, Founder. The one provenance that cannot be traced is his interest in improving methods, making processes work in reality and tracking their goals with proof.
Bachelor of Science in Management and Master of Science in Finance, with almost 10 years of experience leveraging technology for businesses.
Started improving processes from the first job, an internship in the finance department of one of the largest German engineering companies in the world, and never stopped since: moved into enterprise consulting to do exactly that, delivering projects in data science, process mining, cloud architecture and automation.
Most recently led finance and operations across the brands of one pharmaceutical group, each with its own processes, deepening the understanding of what highly regulated industries demand.
Pre-audit
The pre-audit is free and comes before anything is signed. Rextuvia reviews which systems hold the process, which tables and fields exist, how far back they go and how clean they are, and returns a written view of what a Pilot or a Project could and could not find.
What Rextuvia asks for
Read access or an extract of the tables behind the processes in scope, under an NDA and a data processing agreement. The list of tables is TK and will come from the extraction plan.