How should a small or midsize manufacturer begin digital transformation?

Do not start by purchasing a complete digital-manufacturing stack. The client explains current losses, work practices, and desired business outcome. Wavesteam reviews the site, documents, and available data, then recommends one first loop whose loss is measurable, data is obtainable, and improvement owner is known. Expand only after a complete operating cycle produces evidence of value.

If paper and spreadsheets deliver reliably, the first improvement may be unified codes for materials, equipment, operations, work orders, shifts, and defect reasons plus a measured baseline—not MES. The project should answer a business question such as the downtime caused by shortages, reporting delay that harms scheduling, or the operation producing the most rework. A dashboard without baseline, data owner, and operating response rarely creates a result.

When separating device, connectivity, and platform responsibilities, also compare Can Wavesteam build digital applications on a BIM foundation?; the linked guidance adds context that should be considered in the same decision.

Starting route	Suitable state	Value	Failure risk	Default choice
Standardized paper/shared sheet	Changing process, low-frequency single plant	Align fields, ownership, metrics	Limited access, concurrency, traceability	Use while process remains unstable
Standard inventory/ERP/MES SaaS	Conventional process and few interfaces	Rapid standard order, stock, reporting	Site must adapt to product boundaries	Buy when it covers most critical work
Configurable platform plus interfaces	Existing ERP missing a shop-floor loop	Preserve core and add scan, exception, dashboard	Interface and master-data conflict	Often pragmatic for an SME
Custom production system	Genuine process, pricing, or traceability difference	Fits and preserves differentiated workflow	Higher build, operation, staffing dependency	Prove the difference first

ISA-95 helps assign ownership and exchange between enterprise planning and manufacturing operations without requiring every layer of ERP, MES, and SCADA. A first loop can be work order to shift, scan reporting, exception, and ERP writeback if order, operation, worker, equipment, quantity, and quality reconcile.

For production reporting, map who receives the job, starts work, records accepted and defective output, returns rework, approves a shift, and updates finance or ERP. Every field has a source, maintainer, update time, null rule, and conflict authority. A PLC “run/stop” signal cannot explain setup, shortage, maintenance, or trial; device events must join work order, operation, and human reason codes.

The pilot spans a line or workshop through its real shift and settlement cycle, not a few pages. Before and after, measure reporting delay, late entry, stock variance, downtime-reason completeness, response time, and reporting effort on the same definition. Traceability pilots sample whether a finished item reaches batch, operation, equipment, person, and inspection. Software cannot claim every concurrent production gain.

Expand when operators can complete work at normal takt, managers act on the data, and core metrics improve over a full cycle without extensive offline correction. If these fail, correct master data and process before adding modules. Wavesteam's public factory management case is first-party evidence of relevant direction, not an instruction to copy one architecture.