Edcen Data All articles
Data Governance

Who Actually Owns That Dashboard? Rethinking Analytics Outputs in Britain's Mid-Market

Edcen Data
Who Actually Owns That Dashboard? Rethinking Analytics Outputs in Britain's Mid-Market

Ask a group of operations directors at a Midlands manufacturing business how many dashboards the company runs, and the answers will differ by a factor of three. Ask who owns them and the room tends to fall quiet. That silence is the most dependable symptom of an organisation that has acquired a reporting habit rather than a data product capability.

The distinction is not semantic. A dashboard is an artefact: it is produced, circulated and eventually orphaned. A data product is a small piece of infrastructure with a named owner, defined consumers, agreed quality expectations and a visible cost of operation. Firms that blur the two end up with an estate that grows faster than the decisions it supports. Before approving the next request, it is worth running six questions past the person making it.

The difference that actually matters

A report answers a question once. A product answers a recurring class of questions for a known group of people, and it is maintained accordingly. In lean UK teams — often one analytics lead, two analysts and a part-time data engineer — that distinction determines whether capacity is spent on maintenance or on new work.

A useful shorthand: if nobody would notice the output breaking for a fortnight, it is not a product. It is a snapshot that has outlived its purpose.

First: who is named, and have they accepted the role?

Ownership assigned on a slide is not ownership. The test is whether a specific person — usually a commercial or operational manager, not an analyst — can describe the output, its users and its known limitations without reference to the data team. Where every asset defaults to the analytics function, governance collapses into a queue, and the queue becomes the constraint on everything else.

For mid-market firms, one accountable business owner per product is usually sufficient. A deputy is sensible where the owner travels or sits in a different site.

Second: what does 'working' mean on a Tuesday morning?

Service expectations are where many homegrown products quietly fail. Freshness, availability, acceptable error tolerance and the escalation route when something looks wrong should be written down in plain English — three or four sentences, not a service catalogue.

Proportionality matters. A regional distributor does not need 24/7 monitoring for a stock report read at 8am each weekday. It does need to know that a failed overnight refresh will be spotted before the morning meeting, and by whom. Without that, users develop workarounds, and workarounds become the real system.

Third: will anyone outside the original request use it?

Reuse is the clearest commercial justification for treating an output as a product rather than a piece of analysis. The question is whether a second, identifiable group of users is likely to need the same definitions within six months.

Where the honest answer is no, an ad hoc analysis or a well-governed spreadsheet is often the better answer, and considerably cheaper. Where the answer is yes, the case for building once — with shared definitions and controlled access — becomes straightforward.

Fourth: what does it cost after go-live?

Build cost is the figure that gets quoted. Run cost is the figure that gets forgotten. Licence seats, warehouse compute for refreshes, engineering time for schema changes, governance upkeep and periodic revalidation all recur. A reasonable planning assumption for lean teams is that annual running cost lands somewhere between a fifth and a third of the initial build, though the ratio varies with complexity and source volatility.

Asking the question changes behaviour. Requests that survive a cost-of-ownership conversation tend to be the ones worth building.

Fifth: is the governance proportionate to the risk?

Not every product needs the same controls. A margin report drawing on commercially sensitive pricing warrants tighter access than an internal operations view. The minimum for any product is a documented definition of its key measures, a record of source systems, an access model and a retention position.

Where personal data is involved, UK GDPR obligations apply to the output as much as to the source. Data protection by design, as set out by the Information Commissioner's Office, is easier to evidence when the product has a short written record from the outset rather than a retrospective tidy-up.

Sixth: which decision moves, and by what amount?

The question that separates products from ornamentation. A credible answer names the decision, the person who takes it, the current baseline and the direction of expected change. Precision is not required; direction and rough magnitude are.

If the reply is that the dashboard gives 'better visibility', the request is not yet ready. Visibility is an input to a decision, not evidence that one will improve.

Putting the six to work

Applied as a gate, these questions are quick — a short conversation, not a business case. Applied periodically across the existing estate, they form the basis of a retirement list, which is often the fastest route to capacity in a lean team.

A quarterly review with the finance or operations director, using the same six headings, keeps the portfolio honest. The aim is not to discourage requests. It is to make sure that each one leaves behind an asset somebody is prepared to stand behind.

All Articles

Related Articles

The Decision-First Agenda: How UK Businesses Can Make Data Strategy Concrete

The Decision-First Agenda: How UK Businesses Can Make Data Strategy Concrete

The Boardroom Data Audit: Ten Checks Before AI Gets the Green Light

The Boardroom Data Audit: Ten Checks Before AI Gets the Green Light

Beyond Gut Feel: A Finance-Proof Way to Value Data Work in 2026

Beyond Gut Feel: A Finance-Proof Way to Value Data Work in 2026