Vertesa. An open case, with the data
Vertesa is a fictional precision mechanics company: 120 people, three products, three lines, twenty-four suppliers. We use it throughout the series because no real client data can appear here. We publish its synthetic data, five tables with eleven hidden problems, and the exercises to find them. For courses, districts and management meetings.
Part of the program The invisible Arsenal
- Sector
- Precision mechanics
- People
- 120
- Products
- VX-200 valve body, AL-40 linear actuator, GR-12 gearbox
- Lines
- Three, on two shifts
- Suppliers
- 24, in seven countries
- Period
- April to June 2026, extracted on July 31, 2026
Five CSV files, as they come out of an ERP
Semicolon separator, UTF-8 encoding, decimal comma, dates as year-month-day (the column names are in Italian, as in the original). They open in any spreadsheet. They are as we found them in real companies: with the empty fields, the free text and the inconsistencies left in place.
Suppliers
24 rowsMaster data of the twenty-four suppliers: country, category, single source, declared lead time, qualified alternative, date of the last audit.
Columns: id_fornitore · ragione_sociale · paese · categoria · fonte_unica · tempo_consegna_dichiarato_gg · alternativa_qualificata · ultima_verifica · referente_acquisti · note
Download the CSV vertesa-fornitori.csvBill of materials
60 rowsSixty components across the three products: supplier, country of origin, unit cost, quantity per product, criticality.
Columns: codice_componente · descrizione · famiglia · prodotto · nome_prodotto · id_fornitore · paese_origine · costo_unitario_eur · quantita_per_prodotto · criticita · note
Download the CSV vertesa-distinta-base.csvNon-conformities
96 rowsThe quarter’s log: free-text description, cause when there is one, severity, status, decision date and corrective action.
Columns: id_nc · data_apertura · reparto · prodotto · codice_componente · descrizione · causa · gravita · stato · data_decisione · giorni_alla_decisione · azione_correttiva · responsabile
Download the CSV vertesa-non-conformita.csvOrders and deliveries
140 rowsPurchase orders since January: promised and delivery dates, ordered and delivered quantities, status.
Columns: id_ordine · id_fornitore · codice_componente · descrizione · data_ordine · data_promessa · data_consegna · quantita_ordinata · quantita_consegnata · stato
Download the CSV vertesa-ordini-consegne.csvProduction
372 rowsThe production log by day, line and shift: nominal parts, produced parts, scrap, downtime minutes and downtime cause.
Columns: data · linea · turno · prodotto · pezzi_nominali_turno · pezzi_prodotti · scarti · pezzi_buoni · fermo_min · causa_fermo
Download the CSV vertesa-produzione.csvDownload everything (ZIP)
The five tables and the readme file in one archive.
692 rows · v0.1
No published solution
Hidden in the data are eleven problems we have seen in real companies, anonymized and rebuilt on Vertesa. We do not list them: they are the exercise. They belong to five families: the same supplier written in several ways, causes and downtime in free text, dates and quantities that do not add up, declared and real capacity that diverge, component origin missing where it matters.
They are all problems no software solves on its own, and that one afternoon of work with the data in hand makes visible. It is the same afternoon in which, in the series, Vertesa writes its management charter and its non-conformity taxonomy.
Twelve exercises, three per function
Every function works on the same tables and reaches a result the others can check. Management closes with the customer’s question.
Purchasing
- 01Build the single-source supplier map: one row per critical component with country, declared and real lead time, qualified alternative.
- 02Compute the non-EU cost share of each product and compare it with EDIP’s 35%: which product fails, and which data is missing to be sure?
- 03Count the suppliers: how many are there really? Which ones have a last audit older than two years and no qualified alternative?
Quality
- 01Read the non-conformity descriptions and propose a closed list of causes: how many are there? Which is the most frequent, and why does it look like the least visible in the log?
- 02Measure the cycle between opening and decision: median, how many have been open for more than ninety days, how many have a corrective action that says something.
- 03Reclassify the descriptions with an AI tool using your taxonomy and compare the result with the group: where do you disagree, and who decides?
Production
- 01For each line, compare declared capacity (nominal per shift, times shifts, times days) with the real one for the quarter: which line diverges most, and through which mix of yield, scrap and downtime?
- 02What share of downtime minutes has a cause that says nothing? What would you change in the logging form on Monday?
- 03With the 90-day calculator and the orders and bill-of-materials data, estimate how many AL-40 actuators Vertesa can deliver: which constraint binds?
Management
- 01Answer the customer’s question within an hour: “how many AL-40 in ninety days, and where do the components come from?”. Write the answer, the assumptions and the data you were missing.
- 02Fill in the short-cycle checklist for non-conformities: who reports, who decides, in how many days, with which data.
- 03Decide the six data points to keep up to date to answer within an hour next time, and assign each one an owner and a place where it lives.
In class, in the company, with AI
In class, two hours
Twenty minutes to read the tables, one hour in groups split by function, forty minutes in plenary with the customer’s question. It works for production management, quality and purchasing courses and for district training.
In the company, one afternoon
The heads of purchasing, quality and production work on Vertesa with the same data. Then they repeat the same exercises on their own data: the distance between the two experiences is the measure of what is missing.
With an AI tool
The data is synthetic and internal-class: it can be loaded into any tool your management charter allows. Ask the model to group the non-conformity descriptions, then compare the result with the taxonomy the group wrote. The model proposes; quality decides the taxonomy.
The guide with the problems and the solution paths is available on request
We do not publish it because otherwise the exercise would not exist. Write to us naming the course, district or company where you will use the case, and you will receive it in an open format. No cost, no commercial proposal.
Open data, generated with a fixed seed
- The data is released under the Creative Commons Attribution 4.0 license: it can be used, modified and redistributed citing “Yempik, Vertesa open case v0.1”.
- The files are generated by a script with a fixed seed: same seed, same files. Later versions will change the version number and be noted in the program changelog.
- Vertesa does not exist. No data, document or name of a real client appears in the tables; any resemblance to real companies is the resemblance among all the SMEs we have met.