Digital maturity model & Diagnostic
Digital transformation is not a single feature: it is a path that advances in stages.
Variable #1 (the unknown that dominates everything): the shop-floor / ERP data spine (C2) - whether machine, order and quality data resolve into one record, or live in per-line silos reconciled by hand. It gates everything above and is the priority question.
The model rests on recognized, validated standards (not invented). This is an automated preliminary screening from public information only - the observations are hypotheses to confirm together.
1Our role: architects, not feature vendors›
| A feature vendor sells you... | AIXEL - architect |
|---|---|
| "here is an MES / here is a dashboard" | "here is how your whole operating and data system should look, and how to get there in sequenced waves" |
| sells a product | sells a destination map and the guidance along it |
| starts from a feature | starts from the maturity level and the honest gap |
You already run real production with real data being generated on every line. Today that data is not unified. When it is, it becomes the defensible asset that lets you scale margin, not just volume.
Market context. The sector leaders now compete on data and connected operations, not headcount. A unified data layer is what lets a manufacturer grow while holding margin.
2The reference: target architecture›
Six layers bottom-up. The upper layers do not function without the lower ones: that is the whole logic of the sequence.
The sector's connected-operations leaders run a unified data layer feeding predictive maintenance and real-time margin. For an SMB these are aspirational: they define the ceiling (L4). The realistic near-term destination is Connected (L2).
3The maturity ladder L0-L4›
The five rungs are anchored to recognized frameworks (the standard column), validated on the primary source.
| Lvl | Name | What it is | ≈ Acatech I4.0 |
|---|---|---|---|
| L0 | Fragmented | Siloed tools / manual; no unified data; paper or spreadsheets on the floor | Computerisation |
| L1 | Standardized | Unified ERP/MES core; electronic documentation; digital basics across lines | Connectivity |
| L2 | Connected | Live data layer; one record across machines/orders/quality; a 360 view | Visibility |
| L3 | Intelligent | Analytics/AI embedded in decisions; predictive maintenance; managed BI | Transparency / Predictive |
| L4 | Predictive / Platform | Predictive support in production with model-risk controls; each new line plugs in | Adaptability |
The ambitious capabilities sit at L2-L4, but they rest structurally on L1 (a unified core + structured data). The honest first question is not which feature you want, but which rung you are really on.
4Where you are now: maturity map›
current state (hypothesis) target to be defined
| Dimension | L0 | L1 | L2 | L3 | L4 | Confidence | Dist. |
|---|---|---|---|---|---|---|---|
| C1 · Operating core (ERP/MES) | Medium | 1 | |||||
| C2 · Data & interoperability (single source of truth) ★ | to be defined | ~1.5 | |||||
| C3 · Access & engagement (customer front door) | High | 1 | |||||
| C6 · AI & automation (spine) | Low | ~1.5 | |||||
| O1 · Digital strategy & leadership | Medium | 1 |
Weighted composite: L0.8 → L2.0 · weighted total gap ~1.2 rungs, concentrated on the gating foundations (C1, C2). Reported as a hypothesis, not a grade.
A real operation generating real data, with a genuine strength in the customer front door (C3), sitting on foundations of unknown, probably-low integration (C1/C2). The honest destination is Connected (L2): unify the data and prove margin - not AI-native. The gap is modest and foundation-shaped.
5 · The dimensions: levels, questions, reading
C1Operating core (ERP/MES)weight 3 · L1›
Questions
- What single system, if any, does a job flow through from order to delivery to invoice?
- Are the lines/sites on one ERP/MES or separate systems?
- Can you show the tool where a live job is managed, and export one job end-to-end?
Probably L1. Digital operating systems exist, so above L0 - but no evidence of a single core spanning all lines.
C2Data & interoperability (single source of truth)weight 3 · to be defined›
Questions ★
- When you run an order, do machine, order and quality data live under one record or in separate tools?
- How is a job's data reconciled across the shop floor, the ERP and quality?
- Is there a data model / systems diagram? Can you export one job end-to-end?
One of the two decisive unknowns. No integration platform observable; machine, order and quality data likely live in silos. Gating + low confidence: no level asserted.
C3Access & engagement (customer front door)weight 1 · L2›
Questions
- How many orders actually come through the online flow vs by phone/email?
- Is the customer portal used, and for what?
- Can you share online-order volume and portal active-user stats?
Probably L2 (High). A working quote/order flow and an account portal were observed - a stronger front door than most SMB manufacturers have. Credit it.
C6AI & automation (spine)weight 1 · L0-L1›
Questions
- Where is AI used today, and who checks the output?
- Which processes are the most manual and repetitive (automation candidates)?
- Are there internal policies on AI and data?
Probably L0-L1 (Low). Staff plausibly use gen-AI tools ad hoc; no automation platform or governed AI workflow observable.
O1Digital strategy & leadershipweight 2 · L1›
Questions
- Who owns your own data/systems (not customers'), and is there a plan to improve them?
- Is there a digital roadmap with a budget line?
- How are internal digital initiatives funded?
Probably L1 (Medium). No visible owner or roadmap for the company's own data/digital maturity; digital shows up as isolated projects, not an operating capability.
6From diagnostic to roadmap: waves, economics, risks›
The distance from the model is your roadmap. Priority follows the dependencies: first what unblocks the rest (core + data layer), quick wins in parallel.
See behind the floor
- Resolve the C1/C2 unknowns: systems and data-record inventory
- Deliverable = the true current state
The data layer
- Converge on one ERP/MES core buy
- Stand up one record across machines/orders/quality build
Connect & measure
- OEE + margin-per-line dashboards integrate
- CRM on the customer base buy
Intelligence
- Predictive maintenance (governed) build
- Planning automation at the seams
Build / buy / integrate
| Step | Choice | Rationale |
|---|---|---|
| ERP/MES core, CRM, BI, dashboards | buy | commodity best-of-breed; do not build |
| The unifying data layer + shop-floor integration glue | build | the differentiating connective tissue - the architecture, not a tool |
| AI/predictive workflows | integrate | compose off-the-shelf models into governed workflows |
Anti-patterns to avoid: (1) buying one MES and calling it the platform - the value is the integration; (2) building the core from scratch - buy a proven one.
Cost-of-current-state levers, to model on your data: unplanned downtime = hours x loaded line rate; scrap/rework = rate x volume x cost; manual reconciliation = hours/week across tools. Wave 1 (the data layer) is the enabler that all later waves stand on. (illustrative, to model on your data - no prices)
Key risk: the data foundation (C2) is unknown - the whole roadmap is provisional on it, so Wave 1 starts with an inventory before any build. Any predictive/quality AI needs a validation basis before it touches production.
7 · Interventions you can start now
In parallel with the roadmap, three concrete interventions that touch little or none of the deep foundations, are independent of Variable #1, and each seed a real wave.
| Criterion | A . Dashboard | B . Capture | C . AI quoting |
|---|---|---|---|
| Independent of the data unknowns | high | high | high |
| Visible value | high (margin view) | high (leak found) | high (throughput) |
| Roadmap wave | Wave 2 | Wave 1 | Wave 3 |
The fastest way to make this concrete is a working session on one real line: we show where margin leaks and map the first wins together.
Book a working session8Confidence, limits, methodology & sources›
This is an automated preliminary screening from public information only (the website + independent web research). Every current-state observation is a hypothesis with a confidence tag, and each claim traces to a source. The reference model rests on recognized, validated standards.
Limits. We could not see behind the shop floor: the operating core (C1) and the data layer (C2, the Variable #1) rest on inference and absence of public evidence - so they are left to be defined and are the priority questions. The assessment is a point-in-time snapshot.
Additional references and the market-vendor map available on request, to support the build/buy/integrate choices.