The Digital Compass: Steering Instrument or Statement of Intent?
By the end of 2027, generative AI is meant to be available at every standard workstation in the Schleswig-Holstein state administration. The tool meant to track this and dozens of other deadlines across the state is itself only being piloted in three agencies from 2026.
Schleswig-Holstein has updated its Digital Strategy 2026, a substantial strategy document with a stated ambition to be a frontrunner. It contains an announcement that is easy to overlook because it sounds technical: not until 2027 is the project management software Clarity meant to be used, as a rule, as the state-wide, central steering tool (p. 70). Until then, pilot projects will run in three agencies first. This is notable, because this very instrument is meant to ensure in future that the many other deadlines in the same document are kept — several of which fall due before or alongside this pilot period. An administration is announcing what it wants to be steered by in future before the steering itself is in place.
What the document is — and what it is not
The Digital Strategy Schleswig-Holstein 2026 was issued by the State Chancellery, dated March 2026. It is the update to the “Digital Compass”, first adopted by the cabinet in December 2022 and further developed in 2024. According to the document, the Compass serves, together with the Digital Strategy, as the binding frame of reference for all strategies with a digital dimension in the state (p. 18). Its vision is correspondingly broad: “We use digital transformation to develop Schleswig-Holstein into a leading digital region and to shape a sovereign, digital society” (p. 17). Five consolidated guidelines — from the digitalisation of the administration through data-based administrative action to digital innovation — are meant to give this ambition concrete form in the day-to-day work of public authorities (p. 17 f.).
The document is structured into Part A, covering the technical and organisational foundations — sovereignty, cloud, data, AI, organisation, steering — and a new Part B, which brings together sector-specific projects ranging from healthcare to fisheries and places societal benefit at the forefront. Opening up to actors outside the state administration, it states, is crucial for increasing the impact and acceptance of the strategy (p. 23). A separate Part C, setting out the long-term target vision of a “Digitally Cooperating State”, rounds off the paper.
What matters for the assessment is what the paper is not. It is not an empirical survey with a sample size and margin of error, but a political strategy document. Each of the roughly twenty vision passages describes a target state, not a measured current state. No independent evaluation of the preceding 2023 strategy is referenced in the text, and no sanction or escalation mechanism for missed deadlines is named. Anyone looking to this document to find out how far digital transformation in Schleswig-Holstein has actually progressed will find target markers — not a position fix. That is not a shortcoming of the document but its genre: a strategy sets out direction and deadlines; it does not measure itself.
Five key findings from the paper
First: every one of the numerous subject areas follows the same pattern. A long-term vision with a time horizon of five to ten years sits alongside short- to medium-term, concretely dated individual targets, mostly with target years between 2026 and 2028. This structural feature runs through the entire document, for instance on digital sovereignty (p. 28), the sovereign cloud (p. 40), and data-based administrative action (p. 46). The gap between a broadly framed ambition and a dated individual measure is therefore not an isolated case but the document’s recurring blueprint.
Second: concrete “lighthouse” projects are meant to operationalise the vision, each with its own deadline. The Open Source Program Office SH is to be in place by mid-2027 (p. 30), the Competence Centre for Data SH (KoDa.SH) as early as the end of 2026 (p. 48), the Central Grant Management SH, after piloting by mid-2026, is to be fully rolled out by the end of 2027 (p. 91), and the Digital Information System for Civil Protection is planned to run for at least until 2030 (p. 83). Each of these projects has its own sponsoring body, its own deadline, and its own section on added value — the craftsmanship in the detail is unmistakable.
Third, on AI: by the end of 2027, generative AI is meant to be available at every standard workstation in the state administration (p. 50). In parallel, an AI governance framework is meant to be developed only by the end of 2026, defining legally sound and ethical guidelines for its use (p. 50). The rule therefore does not precede the rollout in time; it runs only around a year ahead of it — while the rollout itself already begins in the same year the governance framework is being developed.
Fourth, and this is the core of this article: strategic steering itself is the least developed element in the entire paper. Clarity is meant to provide a central dashboard for digitalisation projects that makes progress and resource use transparent across the state (p. 70) — that is the ambition. The timeline for this: pilot projects for a strategic project portfolio management approach are first to be built up in three state agencies by 2026 (p. 70). According to the document, the experience from these is likewise only to be systematically evaluated by 2026, and a state-wide rollout is envisaged, as a rule, only from 2027 — and even then initially binding only for cross-departmental projects, with use remaining optional for purely departmental ones (p. 70).
Fifth: according to the document, project portfolio management is meant to be interlinked with the IT budget so that funding allocations are tied to strategic goals. The digital strategy is meant to support efficient and transparent prioritisation of IT and digitalisation measures in budget planning, particularly in times of budget constraints (p. 72). This mechanism too — the financial enforcement power of the strategy — hangs on the same timeline as the steering instrument whose pilot phase is only just beginning, and is, according to the document, oriented to the annual update of the overall IT plan (p. 72), without a specific year being named from which the interlinking actually takes effect.
Three further observations
The same pattern of an ambitious figure and a staggered deadline also shows up in data-based administrative action. By the end of 2026, a central system for metadata registration is meant to be built, in which, by the end of 2027, at least 80 per cent of the state administration’s relevant data holdings are registered and provided with standardised access routes (p. 47). An automated reporting and monitoring system is meant to deliver at least 70 per cent of regularly produced reports faster and error-free by the end of 2028 (p. 47). Both figures are formulated concretely and verifiably — progress compared with the more qualitative vision statements in other subject areas. At the same time, the same caveat applies here: whether these 80 and 70 per cent figures are actually achieved can only be measured once the portfolio management meant to track such target achievement is itself beyond the pilot phase.
As background to the steering ambition as a whole, it is worth looking at Part C: the update culminates there in a target vision called the “Digitally Cooperating State”, under which the administration no longer wants to steer by input, such as the number of applications, or by output, such as processing time, but by societal impact — specifically, whether the lives of citizens have noticeably improved (p. 102). This is a demanding shift in the steering logic: away from easily countable process metrics, towards an impact that can only be established with a time lag and with indicators specifically developed for the purpose. Precisely such a shift needs a robust steering instrument sooner rather than later — because, unlike simple processing times, impact metrics cannot be read off from existing case-handling systems but first have to be built up through the planned portfolio management and the accompanying data projects.
The target vision also sets a high bar at the level of service experience. Citizens are meant to be able to interact with the administration in future as easily as with modern apps, and benefits are meant to be granted proactively — the document cites automatic child-benefit approval following a birth as an example (p. 102). Such a proactive ambition presupposes that data flows reliably across case-handling systems and departmental boundaries, and that the underlying processes are already documented and standardised cleanly enough today for automated decisions to be derived from them. This ambition, too, therefore depends on the same foundational projects — data standards, metadata registration, interoperability — whose own deadlines extend to 2027 and 2028. The further the ambition for the service experience reaches, the more weight falls on the question of whether the foundational work is in place in time.
What is new about this
That digitalisation strategies link visions to individual measures is, in itself, not a new observation. I have described in earlier articles why digitalised processes without accompanying organisational development only make their old weaknesses visible faster, and why new organisational charts alone change nothing as long as the work underneath them does not change. What is new about this document is that it is a concrete strategy paper from a state administration, against which this criticism can no longer merely be asserted in general terms but traced in detail — with real deadlines, real responsibilities, and a named, but itself still unfinished, steering instrument.
The 2026 update goes explicitly further than the preceding 2023 version: it supplements the technical and organisational foundations with its own societal section covering sector-specific projects (p. 75) — a broadening that makes the question of steering depth all the more pressing. The more departments with their own lighthouse projects are brought under a shared umbrella, the more depends on a portfolio management approach that does not merely list these projects but actually makes them comparable and steerable. A strategy that broadens without a steering instrument growing alongside it widens the gap between ambition and demonstrability rather than closing it.
Michael’s position
A vision plus a long list of lighthouse projects is not yet steering. Implementation measures only become more than a statement of intent through named responsible parties, binding timeframes and — as this document clearly shows — a steering instrument that actually works. This is precisely the standard the Digital Compass must be measured against. What stands out is that the paper essentially names the risk itself, without framing it that way. It sets the binding deadlines for operational projects in time ahead of the binding use of the instrument meant to make their observance transparent. By the time a project such as KoDa.SH falls due at the end of 2026, state-wide portfolio management will not yet exist in the field at all — it will still be in pilot operation in three agencies. Whether a stalling lighthouse project would even become visible before its deadline passes remains an open question in the document.
This is not a criticism of the craftsmanship of the paper — the subject areas are well thought through, the dependencies between them clearly stated, the societal ambition genuinely meant. Every subject area explicitly lists “dependencies and synergies with other subject areas”, for instance noting, for project portfolio management, that it is closely interlinked with the IT budget and makes priorities financially effective (p. 71). This care in cross-referencing makes the gap in the temporal sequencing all the more striking: the dependencies between the subject areas are well thought through, but the dependency of the subject areas on a steering instrument that is not yet working state-wide is not identified as a risk in its own right.
The objection therefore concerns the sequencing, not the intent. A steering instrument that only takes full effect after the first operational deadlines cannot deliver, in the decisive phase, what it is meant to deliver. Only once the announced project portfolio management actually draws consequences for stalling projects — and does so before, not after, the first deadlines have passed — will the vision become a steering instrument, rather than merely another chapter in a well-written strategy paper. A state administration that only builds its own progress monitoring alongside the first implementation deadlines leaves to chance what steering should actually be responsible for: whether a delayed project is noticed in time, or only once the deadline has already passed.
What this means in practice for public authorities
First: anyone announcing a steering instrument should have it operational before the first operational deadlines, not build it up in parallel with them. Otherwise, the most ambitious projects run without oversight precisely in the phase in which they are most likely to stall — namely at the start, when responsibilities are not yet established and resources not yet reliably allocated.
Second: a portfolio management approach is only worth as much as the consequence mechanism behind it. An administration that introduces a dashboard for transparency but does not define what happens when a deadline is flagged red has created visibility, but not steering. Transparency is a precursor to steering, not steering itself.
Third: the interlinking of strategy and budget, as announced here for the time horizon of the overall IT plan, belongs at the start of a project, not in its middle. Where funding allocation and strategic priority are only brought together afterwards, the first budget decisions have already been made without that filter — and are difficult to correct retrospectively without jeopardising projects already under way.
Fourth: other state and local authorities drafting or updating their own digital strategies should check the sequencing before publishing deadlines. A vision with a date is quickly formulated. An instrument that tracks whether it is met typically takes longer than a cabinet meeting — and should therefore come first, not last. An authority that places its steering instrument at the end of the list of subject areas, as happens here under the heading “Strategic Steering”, should not also place it last organisationally in implementation.
Sources
State Chancellery of the State of Schleswig-Holstein (Minister-President of the State of Schleswig-Holstein): Digitalstrategie Schleswig-Holstein 2026 – Digitale Verwaltung. Gesellschaftliche Wirkung. (Digital Strategy Schleswig-Holstein 2026 – Digital Administration. Societal Impact.), dated March 2026. Available at schleswig-holstein.de/digitalstrategie.