Digitalisation is routinely marketed as the straightest line from the shop floor to higher productivity. Install the software, connect the machines, automate the paperwork, and competitiveness should follow. Yet inside real manufacturing firms, especially mid-sized ones operating under tight labour markets and uncertain demand, digital transformation rarely behaves like a switch you flip. A recent pilot mixed methods study of Croatian metal and electrical manufacturers offers a more realistic picture: many firms are adopting elements of Industry 4.0, but the performance story depends less on “having technology” and more on whether firms can integrate it into everyday workflows under severe organisational constraints.

The study is positioned as a rigorous pilot. It does not claim causal impacts or use administrative productivity accounts; instead, it examines how managers perceive two outcomes that matter strategically: productivity-based competitiveness in foreign markets and expected growth. This choice is not a weakness but an honest reflection of the evidence many policy programmes actually have at hand. In early-stage transitions, perceptions often capture managerial priorities, bottlenecks, and readiness for scaling, while objective performance data can lag digital investments by years.

To avoid the usual problem of vague “digital maturity” labels, the study builds a transparent measurement system directly from item-level survey responses. It constructs a capability-focused digital maturity index from four pillars: digital infrastructure breadth (systems and tools), production digitalisation practices, R&D process digitalisation practices, and integration intensity (whether systems talk to each other). It also builds separate indices for strategy–skills integration, barrier severity, and AI readiness/adoption. A crucial technical complication is that some practices are not relevant for every firm. Managers frequently answered “No need” to particular items, especially in the R&D block, which would normally force researchers to delete firms and shrink the sample. Instead, the study uses denominator-aware scoring: each firm’s index is computed using only the items that are applicable to that firm, and the analysis is re-run under a conservative alternative that treats “No need” as non-adoption. This matters because measurement choices can drive conclusions in small samples.

The headline quantitative message is sobering for technology evangelists: in this pilot, higher digital maturity does not clearly translate into higher perceived competitiveness through labour productivity. In plain terms, firms that score higher on the combined digital maturity index do not systematically report stronger productivity-based competitiveness. The relationship remains weak even when the index is recalculated under the conservative coding rule. Put differently, the data do not support a simple story in which more digital capability, on its own, reliably produces a better competitiveness perception within this group of firms.

For expected growth, the picture is slightly more encouraging but still cautious. Firms with higher digital maturity tend to show a somewhat more positive outlook, but not in a way that allows strong conclusions in a pilot sample. The direction is consistent with the intuition that digital capability correlates with ambition and confidence, yet the evidence is not definitive enough to claim a robust pattern.

Strategy is often invoked as the bridge between technology and outcomes. Here, the study again finds that formal strategy and role-based digital skills, captured through a strategy–skills integration index, do not show a strong relationship with perceived competitiveness or growth. Nor does strategy–skills integration appear to strengthen the link between digital maturity and perceived competitiveness. A plausible interpretation, supported by managers’ qualitative comments, is that strategy in many firms is informal, implicit, or frequently displaced by day-to-day operational pressures. Skills constraints also seem less about basic IT literacy and more about time, process capacity, and the difficulty of retaining people who can carry implementation projects through.

The most interesting quantitative signals appear when the analysis looks beyond “digital maturity as a bundle” and examines integration and R&D digitisation more closely. Integration intensity and R&D digitisation show patterns that are suggestive for growth expectations, but they do not deliver a strong, consistent link to productivity-based competitiveness in this pilot. That may sound disappointing, but it is also informative. It suggests that in early-stage digital transitions, integration and digital engineering practices may influence planning horizons, flexibility, and investment appetite before they show up as perceived productivity advantages. It also underscores a familiar theme in the Industry 4.0 literature: capabilities are complementary, and partial adoption can create “digital islands” that deliver local benefits without shifting firm-level performance perceptions.

Where the evidence becomes sharper, and more policy relevant, is barriers. The barrier index is clearly aligned with digital maturity in the expected direction: firms reporting more severe organisational and human-capital constraints tend to have lower digital maturity. This is not a minor nuance. It shifts the implied mechanism from “digitalisation yields productivity” to “constraints prevent coherent digital capability formation.” In practical terms, barriers are not merely dampening the returns to maturity; they are associated with whether maturity emerges at all. That insight has immediate implications for support programmes: subsidising tools or encouraging strategy documents may be less effective than strengthening the implementation capacity that allows firms to integrate systems and sustain change.

The qualitative comments bring this constraint-first story into focus. Managers frequently describe digitalisation as opportunistic and project-based rather than strategy-led. They often talk about fragmented systems and a lack of end-to-end data flow. Labour shortages and the scarcity of experienced technical staff recur as binding limitations. Some firms report using policy instruments such as “digitalisation vouchers” to develop digital transformation strategies, but they also note that strategy documents can remain detached from operational execution. Alongside these digital themes, managers repeatedly highlight administrative and regulatory friction, slow procedures, compliance burdens, parafiscal charges, and delays that slow investment and expansion. Taken together, the comments align with the quantitative finding that “more technology” is not the binding constraint; implementation capacity and organisational frictions are.

AI readiness is treated as exploratory, but it adds a forward-looking angle. Firms that score higher on AI readiness/adoption tend to show a more optimistic growth outlook in descriptive patterns, even if the pilot evidence cannot support strong claims. The likely interpretation is that AI readiness is not only about deploying tools; it may reflect broader managerial orientation toward experimentation, process redesign, and future capability building, attributes that are plausibly tied to growth expectations.

What does all this mean for the productivity debate? First, it is a correction to technological determinism. The pilot does not support the idea that adoption of digital components automatically and quickly yields perceived productivity-based competitiveness. Second, it provides a more credible mechanism: digital capability formation appears to be constrained by organisational and labour frictions, and integration is the hard part. Third, it offers an unusually practical contribution: a transparent measurement architecture that can be replicated and scaled to larger samples and then linked to objective outcomes such as sales per worker, delivery reliability, defect rates, or export participation.

The policy lesson is straightforward but demanding. If the aim is to raise productivity, support programmes should not stop at tool acquisition or strategy paperwork. They should prioritise integration-focused investments that improve data flow across engineering, production, and management; strengthen workforce development and retention mechanisms that sustain implementation; and reduce administrative and regulatory delays that raise the cost of expanding and modernising. Digital transformation is not only a technology agenda, it is an organisational capacity agenda.

BOX: Why Serbian manufacturing should care (even if the sector differs)

Serbia’s manufacturing firms face an EU accession reality: convergence toward EU standards increasingly requires traceability, process reliability, digital documentation, and compliance with evolving ESG and supply-chain requirements. Yet Serbia lacks a comparable, firm-level evidence base on digital maturity, integration, and barriers in manufacturing. The Croatian pilot offers both a warning and a blueprint. The warning is that digital adoption without integration and implementation capacity may not translate into productivity improvements that managers can actually feel. The blueprint is to build transparent firm-level measurement of digital capability and constraints, then scale it to larger samples and link it to objective performance. That evidence would help Serbia target industrial policy toward interoperability, workforce development, and compliance-readiness, the practical foundations for competing inside EU value chains while moving toward EU membership.