On Data Products and the Printing Press

In my experience, it seems far easier for people working in technology and in the many disciplines that enable it to conceptualise and anchor to the systems and platforms that produce data products than to the data products themselves. That imbalance influences the efficacy of data in an organisation and how it is leveraged to create value.

This disposition doesn't seem to appear across comparable contexts. When we think about a book, as consumers, we rarely concern ourselves with the machinery behind it, the printing press, the ink, the typeset. Our attention naturally goes to the book itself, the knowledge it contains, the story it tells or the impact it will have.

We do care about the metadata, its lineage, the author, the language, and its authority on the subject we wish to learn from. In that sense, a book is a data product: Information organised so that we, the reader, can make meaningful use of it. So why is it difficult to instil organisational behaviours that value both the outcome as well as the means, the book as much as the press? How do we ensure that Hannah Arendt’s “triumph of the world of fabrication” applies not only to the logistics of data, but to the substance of what it delivers?

Arendt warned that when the world of fabrication triumphs without reflection, the durable world of meaning gives way to a churn of production, so how do we, as technologists, ensure that the enduring outcome (the data product) of the data press (the data platform) does not simply become silent throughput?

If meaning gives way to production, as Arendt warned, it is often because systems inherit their own logic. We find this logic not only in machines but in accounting models, governance frameworks, and incentives. Each shapes what an organisation can see as valuable.

Limited organisational awareness

In many domains, there is little collective awareness of data as a product or of the architectural rationale for engaging with it as such. Finance, for example, a powerful behavioural driver of how technology is structured and managed, tends to distinguish between Opex and Capex, amortising the latter. It is far more intuitive for Opex to be measured against the press than against the book. Amortisation, an accounting necessity, is undertaken against the large capital investments needed to build out the press itself. The financial logic quietly shapes the cultural one: we fund the creation and maintenance of denormalisation, annotation, fabrication, but only implicitly authorship. By default, we track ROI against capex investment and lose momentum once the project governance boxes are ticked, long before we reach the work of continuously measuring the ROI at the fidelity of a data product.

Emergent product practice maturity

When we look at de facto product practices, concepts like Data Mesh are far from ubiquitous. Creating great products is a tacit exercise, one grounded in hypothesis, design and testing to fulfil jobs-to-be-done for defined personas; those disciplines are difficult to translate to the world of data. It feels more natural to treat the work as a technical endeavour rather than the design and proof loop familiar with traditional product development. The ability to curate a high-quality backlog or canvas that defines a data product in executable terms sits at the intersection of architecture, engineering and product, a space far less codified than that of traditional digital products. Despite the velocity that data has achieved in recent years, particularly against the backdrop of AI, there remains limited industry support for synthesising good data design practices. There are a few good books, but nothing like the volume of influential work that has shaped conventional product design.

Operational gravity in mature enterprises

The enterprise runs on policy, governance, objectives and escalation. Within this context, engineering groups are measured by how well systems are secured, available, automated and maintained. These disciplines define operational excellence and rightly demand focus. Most of this activity happens within the envelope of finite Opex budgets, where every cycle must justify itself against cost and compliance. Unless data products are woven into the organisation’s strategic fabric and recognised within these same operational frameworks, the conversation and the attention will remain dominated by systems.

Celebration of the tangible

It takes a special kind of person to get excited by a demo of a data product or progress towards one. They lack visual engagement to all but a concerned few, and as the most senior stakeholders will have a broad portfolio of responsibilities, it's a niche that is easy to lose engagement on.

It’s a deeply human bias, the comfort of the concrete over the discomfort of the abstract. Where a wireframe, UX or dashboard can capture an audience's attention, a beautifully constructed JSON-LD file with HTML metadata is more likely to make attention drift.

Data Platforms are rarely profit centres, more often cost centres, contributing indirectly to the mission of the company as a second-order effect; it is not the glory work that is celebrated by extroverted Product Owners in town halls.

Overcoming the inevitable

Some of these tensions are inevitable; the systems exist for good reason. The task of the data organisation is to discern what can be changed, what must be accepted, and where influence, that subtle art between control and surrender, can move the indifferent toward the good. As Greenleaf once said, "The danger perhaps is to hear the analyst too much and the artist too little".

The analyst sees the system for what it is: logic, control, measurement, and delivery. Whilst the artist perceives what it could become: a vehicle for change, innovation and growth. Both are necessary, but when the analyst’s voice dominates, the system begins to speak only to itself. The data press hums, yet the books it prints might be read, but are seldom studied or celebrated.

Modern approaches such as Data Mesh seek to restore the artist’s voice. They decentralise ownership, treating data not as exhaust but as authorship, products stewarded within their domains. In parallel, FAIR principles offer a complementary grammar for this practice: to make data Findable, Accessible, Interoperable, and Reusable is, at heart, an act of hospitality. FAIR is the ethic of care that ensures what we produce can be found, understood, and built upon by others, future generations. It is the infrastructure of generosity that allows knowledge to persist beyond the boundaries of its creation.

Together, Mesh and FAIR represent an evolution from platform thinking to product and purpose thinking, from extraction to stewardship, from control to collaboration. They invite technologists to think less like engineers maintaining machinery and more like designers curating meaning: to create data that is not only compliant, but coherent; not only available, but valuable, and they invite both data and its builders to become Product.

In becoming Product, the Data Platform allows us to be both analyst and artist, to perfect the means and honour the end so that even amid the inevitable frictions of systems, transparent influence may still move the indifferent toward the good.